Powerful Forecasting With MS Excel Sample

Readers will be provided a link to download the software
and Excel files that are used in the book after payment.
Please visit http://www.xlpert.com for more information on
the book. The Excel files are compatible with Excel 2003,
2007 and 2010.

Chapter 1: Basic Forecasting Methods p/g 1

Moving Average Forecast a country farm production
Exponential Smoothing Forecast a country farm production
Holts Method Forecast a winter clothing sales
Holts Winters Forecast fishing rods sales

Chapter 2: ARIMA / Box Jenkins p/g 24

Arima Modeling
Subjective Method
Example 1: Daily Sales Forecasting
Objective Method
Example 2: Forecast daily electricity production

Chapter 3: Monte Carlo Simulation p/g 81

Example 1: Coin Tossing
Example 2: Sales Forecasting
Example 3: Modeling Stock Prices

Chapter 4: K Nearest Neighbors p/g 103

Example 1: KNN for classification of home ownership
Example 2: KNN for time series prediction with stock market data
Example 3: Cross Validation With MSE

Chapter 5: Neural Network With MS Excel p/g 117

Example 1: Credit Approval Model
Example 3: Predicting Stock Prices (DJIA)
Example 4: Predicting Real Estate Value
Example 5: Classify Type Of Flowers

Chapter 6: Markov Chain with MS Excel p/g 218

Example 1: Predict delivery time for a courier company

Example 2: Predict brand loyalty of soft drink customers

Chapter 7: Bayes Theorem with MS Excel p/g 231
Example 1: Predict rainy or sunny day

Example 2: Predict stock market trend

Appendix A p/g 239
How to use nn_Solve

Appendix B p/g 241
Simple Linear Regression For Forecasting

Appendix C p/g 251
How to access Excel Solver 2007 and 2010

1

Chapter 1:

Basic Forecasting Methods

Introduction
Forecasting is the estimation of the value of a variable (or set of variables) at some future
point in time. In this book we will consider some methods for forecasting. A forecasting
exercise is usually carried out in order to provide an aid to decision-making and in
planning the future. Typically all such exercises work on the premise that if we can
predict what the future will be like we can modify our behaviour now to be in a better
position, than we otherwise would have been, when the future arrives. Applications for
forecasting include:
- inventory control/production planning - forecasting the demand for a product
enables us to control the stock of raw materials and finished goods, plan the
production schedule, etc
- investment policy - forecasting financial information such as interest rates,
exchange rates, share prices, the price of gold, etc. This is an area in which no one
has yet developed a reliable (consistently accurate) forecasting technique (or at
least if they have they haven't told anybody!)
- economic policy - forecasting economic information such as the growth in the
economy, unemployment, the inflation rate, etc is vital both to government and
business in planning for the future.
Types of forecasting problems/methods
One way of classifying forecasting problems is to consider the timescale involved in the
forecast i.e. how far forward into the future we are trying to forecast. Short, medium and
long-term are the usual categories but the actual meaning of each will vary according to
the situation that is being studied, e.g. in forecasting energy demand in order to construct
power stations 5-10 years would be short-term and 50 years would be long-term, whilst in
forecasting consumer demand in many business situations up to 6 months would be short-
term and over a couple of years long-term. The table below shows the timescale
associated with business decisions.

2

Timescale Type of decision Examples
Short-term up to 3-6 months
Operating
Inventory control,
Production planning,
distribution
Medium-term up to 3-6
moths to 2 years months

Tactical
Leasing of plant and
equipment, Employment
changes
Long-term above 2 years
Strategic
Research and development
Acquisitions and mergers,
Product changes

The basic reason for the above classification is that different forecasting methods apply in
each situation, e.g. a forecasting method that is appropriate for forecasting sales next
month (a short-term forecast) would probably be an inappropriate method for forecasting
sales in five years time (a long-term forecast). In particular note here that the use of
numbers (data) to which quantitative techniques are applied typically varies from very
high for short-term forecasting to very low for long-term forecasting when we are dealing
with business situations. Some areas can encompass short, medium and long term
forecasting like stock market and weather forecasting.
Forecasting methods can also be classified into several different categories:
- qualitative methods - where there is no formal mathematical model, often because
the data available is not thought to be representative of the future (long-term
forecasting)
- quantitative methods - where historical data on variables of interest are
availablethese methods are based on an analysis of historical data concerning
the time series of the specific variable of interest and possibly other related time
series and also examines the cause-and-effect relationships of the variable with
other relevant variables
- time series methods - where we have a single variable that changes with time and
whose future values are related in some way to its past values.
This chapter will give a brief overview of some of the more widely used techniques in the
rich and rapidly growing field of time series modeling and analysis. We will deal with
qualitative and quantitative forecasting methods on later chapters.
Areas covered in this Chapter are:
a) Simple Moving Average
b) Weighted Moving Average
c) Single Exponential Smoothing
d) Double Exponential Smoothing (Holts Method)
3

e) Triple Exponential Smoothing (Holt Winters Method)
These methods are generally used to forecast time series. A time series is a series of
observations collected over evenly spaced intervals of some quantity of interest. e.g.
number of phone calls per hour, number of cars per day, number of students per semester,
daily stock prices
Let
y
i
= observed value i of a time series (i = 1,2,,t)
yhat
i
= forecasted value of y
i

e
i
= error for case i = y
i
yhat
i

Sometimes the forecast is too high (negative error) and sometimes it is too low
(positive error).
The accuracy of the forecasting method is measured by the forecasting errors. There
are two popular methods for assessing forecasting accuracy:
Mean Absolute Deviation (MAD)
MAD = ( | e
i
| )/n => sum of errors divided by the number of periods in the forecast
Where n is the number of periods in the forecast; e
i
= error
Units of MAD are same as units of y
i

Mean Squared Error (MSE)
MSE = ( e
i

2
)/n => sum of squared errors divided by the number of periods in the forecast

a) Single Moving Average
This simplest forecasting method is the moving average forecast. The method simply
averages of the last n observations. It is useful for time series with a slowly changing
mean. Use an average of the N most recent observations to forecast the next period.
From one period to the next, the average moves by replacing the oldest observation in
the average with the most recent observation. In the process, short-term irregularities in
the data series are smoothed out.
The general expression for the moving average is

4

yhat
i
=[ y
t
+y
t-1
+... +y
t-N+1
] / N
How do you choose N? We want the value of N that gives us the best forecasting
accuracy. (i.e. minimizes MAD or MSE). Lets look at an example. Open workbook
TimeSeries in the folder Chapter 1 and goto to Worksheet (Milk Data). Here we have the
daily record of milk production of a country farm.

Figure 1.1
Next, goto Worksheet (ma solution).We will use N = 3, so we enter the formula
=AVERAGE(B6:B8) in cell C9. This formula is simply [ y
t
+ y
t-1
+ y
t-2
] / 3. (see Fig 1.2
below)

5

Figure 1.2
After that we fill down the formula until C25. We take the absolute error at column D and
squared error at column E. See all the formula enter in Fig 1.3 below.
We have the MAD = 2.7 in cell D27 and MSE = 10.7 in cell E27. If at day 20 we have 49
gallons, how do you forecast the production at day 21? To forecast simply fill down the
formula in cell C25 to cell C26 or enter the formula =AVERAGE(B23:B25). Here the
result is 52.3 gallons. (see Figure 1.2)
As you can see with this example, the moving average method is very simple to build.
You can also experiment with N = 4 or 5 to see whether you can get a lower MAD and
MSE

6

Figure 1.3

b) Weighted Moving Average
In the moving averages method above, each observation in the average is equally
weighted whereas the weighted moving averages method, typically the most recent
observation carries the most weight in the average. The general expression for weighted
moving average is
yhat
i
=[ w
t
*y
t
+w
t-1
*y
t-1
+... +w
t-N+1
* y
t-N+1
]

Let w
i
= weight for observation i
w
t
= 1 => The sum of the weights is 1.
To illustrate, lets rework the milk example. Goto Worksheet (wma solution). Using a
weight of .5 enter in cell H2 for the most recent observation, a weight of .3 in cell H3 for
the next most recent observation, and a weight of .2 in cell H4 for the oldest observation.
(see Figure 1.4)
Again using N =3, we enter the formula =(B8*$H$2+B7*$H$3+B6*$H$4) in cell C9
and fill down to C25. (Note: we use the $ sign to lock the cell in H2, H3 and H4).
7

Figure 1.4
The absolute error and squared error are entered in column D and E respectively. Cell
D27 show the MAD and cell E27 is the MSE.
Could we have used a better set of weights than those above in order to have a better
forecast? Let use Excel Solver to check for you. Choose Tools, Solver from the menu.
(See Appendix C on how to access Excel 2007 and Excel 2010 Solver)

Figure 1.5
Upon invoking Solver from the Tools menu, a dialogue box appears as in Fig 1.6 below
8

Figure 1.6

Next we will enter all the parameters in this dialogue box. It will look like Fig 1.7 below
after all the parameters are entered.
Set Target Cell: MAD (D27)
Equal to: Min
By Changing Cells: the weights at H2:H4
Constraints are that the weights that must add up to 1 at H5

Figure 1.7
Click the Solve button. Solver will start to optimize.

9

Figure 1.8
And Keep the Solver solution
Initially we have the MAD = 2.6 in cell D27 (see Figure 1.4 above). After optimizing
with Excel Solver the MAD is = 2.3 and the weights are 0.57, 0.11, 0.32 in cell H2:H4
respectively. (see Fig 1.9 below) So we have improved our model using Excel Solver.
If at day 20 we have 49 gallons, how do you forecast the production at day 21? To
forecast simply fill down the formula in cell C25 to cell C26 or enter the formula
=(B25*$H$2+B24*$H$3+B23*$H$4)). Here the result is 51.4 gallons. (see Fig 1.9
below)

Figure 1.9
10

As you can see with this example, the weighted moving average method is more
complicated to build and will give us better result.
One disadvantage of using moving averages for forecasting is that in calculating the
average all the observations are given equal weight (namely 1/L), whereas we would
expect the more recent observations to be a better indicator of the future (and accordingly
ought to be given greater weight). Also in moving averages we only use recent
observations, perhaps we should take into account all previous observations.
One technique (which we will look at next) known as exponential smoothing (or, more
accurately, single exponential smoothing) gives greater weight to more recent
observations and takes into account all previous observations.

c) Single Exponential Smoothing
Exponential Smoothing is a very popular scheme to produce a smoothed time series.
Whereas in Single Moving Averages the past observations are weighted equally,
Exponential Smoothing assigns exponentially decreasing weights as the observation get
older. In other words, recent observations are given relatively more weight in forecasting
than the older observations.
In the case of moving averages, the weights assigned to the observations are the same and
are equal to 1/N. In exponential smoothing, however, there are one or more smoothing
parameters to be determined (or estimated) and these choices determine the weights
assigned to the observations.
Lets look at Single Exponential Smoothing first. Recall that y
i
= observed value i and
yhat
i
= forecasted value i. The general expression is
yhat
i+1
= oy
i
+(1- o) yhat
i
Which says
forecast for the next period = forecast for this period + smoothing constant * error for this
period
where 0<=o<=1
The forecast for the current period is a weighted average of all past observations. The
weight given to past observations declines exponentially. The larger the o, the more
weight is given to recent observations.
11

So you can see here that the exponentially smoothed moving average takes into account
all of the previous observations, compare the moving average above where only a few of
the previous observations were taken into account.
Dont worry above the equation above. Ill show you how it is easily implemented in
Excel.
Again, this method works best when the time series fluctuates about a constant base
level. Simple exponential smoothing is an extension of weighted moving averages where
the greatest weight is placed on the most recent value and then progressively smaller
weights are placed on the older values.
To start the process, assume that yhat
1
= y
1
unless told otherwise. Do not use this
observation in your error calculations though.
Lets now rework the MILK problem. Goto the Worksheet (exp solution). We use a simple
exponential smoothing with an o of 0.3 as initial value in cell G4.
x
This part of the book is not available for viewing
Please visit http://www.xlpert.com/forecast.htmfor more
information on the book
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
12

It is customary to assume that yhat
1
= y
1
so we enter C6 = B6. After that starting at
period 2 in cell C7 we enter the formula =$G$4*B6+(1-$G$4)*C6 and fill down the
formula until cell C25. (Note: we use the $ sign to lock the cell in G4). This is how the
formula yhat
i+1
= yhat
i
+ o (y
i
yhat
i
) is entered as Excel formula.
The absolute error and squared error are entered in column D and E respectively. Cell
D27 show the MAD and cell E27 is the MSE. We will use Excel Solver to find a better
value for o and also minimize the MAD and MSE.

Figure 1.11
We will enter all the parameters in this dialogue box. It will look like Fig 1.11 above after
all the parameters are entered.
Set Target Cell: MAD (D27); Equal to: Min
By Changing Cells: the weights at G4 Constraints are that the 0 <= G4 <= 1.

Figure 1.12
13

The MAD in cell D27 is equal 2.8 and the o in cell G4 is equal 0.31109. We have made a
little improvement.
If at day 20 we have 49 gallons, how do you forecast the production at day 21? To
forecast simply fill down the formula in cell C25 to cell C26 or enter the formula
=$G$4*B25+(1-$G$4)*C25. Here the result is 51.8 gallons. (see Figure 1.13 below)

Figure 1.13
As you can see with this example, the simple exponential smoothing method is a little
more complicated to build and should give us better result.
Exponential smoothing is useful when there is no trend. However if the data is trending,
we need to use the Double Exponential Smoothing method which will be discussed
below.

14

d) Double Exponential Smoothing (Holts Method)
Double exponential smoothing is defined as Exponential smoothing of Exponential
smoothing. Exponential smoothing does not excel in following the data when there is a
trend. This situation can be improved by the introduction of a second equation with a
second constant, |, the trend component, which must be chosen in conjunction with o,
the mean component. Double exponential smoothing is defined in the following manner :
yhat
i+1
= E
i
+T
i
, i = 1,2,3
where
E
i
= o y
i
+(1-o) (E
i-1
+ T
i-1
)
T
i
= |(E
i
E
i-1
) + (1-|)T
i-1

0< o <= 1 and | is another smoothing constant where 0<=|<=1.
This method works best when the time series has a positive or negative trend (i.e. upward
or downward). After observing the value of the time series at period i (y
i
), this method
computes an estimate of the base, or expected level of the time series (E
i
) and the
expected rate of increase or decrease per period (T
i
). It is customary to assume that E
1
=
y
1
and unless told otherwise, and assume T
1
= 0.
To use the method, first calculate the base level E
i
for time i. Then compute the expected
trend value T
i
for time period i. Finally, compute the forecast yhat
i+1
. Once an
observation y
i
is made, calculate the error and continue the process for the next time
period. If you want to forecast k periods ahead, use the following logic.
yhat
i+k
= E
i
+kT
i
where k = 1,2,3
Open the Worksheet (double exp solution).
The data is for the monthly sales in thousands for a clothing company. Initially we use the
value of 0.2 (cell I4) for o and 0.3 (cell J4) for |. For E
1
= y
1,
we enter =B6 in cell C6.
Then we enter the formula =$I$4*B7+(1-$I$4)*(C6+D6) in cell C7 and fill down to C29.
As you can see from the formula, we start at period 2 in cell C7. (Note: we use the $ sign
to lock the cell in I4). This is how the formula E
i
= o y
i
+(1-o) (E
i-1
+ T
i-1
) is entered
as Excel formula.
After that, we will calculate T
i
= |(E
i
E
i-1
) + (1-|)T
i-1
. It is common practice to enter
0 at period 1 i.e cell D6. Starting at period 2, in cell D7, we enter the formula =$J$4*(C7-
C6)+(1-$J$4)*D6 and fill down to D29. This is how the formula T
i
= |(E
i
E
i-1
) + (1-
|)T
i-1
is entered as Excel formula.
15

Now we combine the 2 formula together and enter the formula =C6+D6 in E7 and fill
down to E29 i.e the formula yhat
i+k
= E
i
+kT
i
. The absolute error and squared error are
entered in column F and G respectively. Cell F31 show the MAD and cell G31 is the
MSE. We will use Excel Solver to minimize the MAD and to find a better value or
optimize for o and |. (see Figure 1.14 below)

Figure 1.14
Invoke Excel Solver to minimize the MAD. We enter all the parameters in this dialogue
box. It will look like Fig 1.15 below after all the parameters are entered.
Set Target Cell: MAD (F31)
Equal to: Min
By Changing Cells: the weights at I4:J4
Constraints are that the 0 <= I4 <= 1, 0 <= J4 <= 1
16

Figure 1.15

Figure 1.16
And Keep the Solver solution.
The MAD in cell F31 has been minimize and is equal to 140.91 and the o in cell I4 is
equal 0.126401. The | in cell J4 = 1. We have made a little improvement after optimizing
with Excel Solver.
If at month 24 we have 2,850,000, how do you forecast the production at month 25? To
forecast simply fill down the formula in cell E29 to cell E30 or enter the formula
=C29+D29. Here the result is 3,037,810. (see Figure 1.17 below)
If you want to forecast k periods ahead, use the following logic.
yhat
i+k
= E
i
+kT
i

In this example we want to forecast sales at month 32 i.e. 8 months ahead, so we enter
=C29+8*$D$29 as you can see in cell E37. The result is 3,868,383. (see Figure 1.17)
E
i
= C29
17

T
i
= D29
k = 8

Figure 1.17
As you can see with this example, the double exponential smoothing method is a little
more complicated to build and should give us better result. What happens if the data
show trend as well as seasonality? In this case double exponential smoothing will not
work. We need to use the Triple Exponential Smoothing method which will be discussed
below.

e) Triple Exponential Smoothing (Holt Winters Method)
This method is appropriate when trend and seasonality are present in the time series. It
decomposes the times series down into three components: base, trend and seasonal
components.
18

Let s
i
= seasonal factor for period i
If s
i
= 1, then season is typical
If s
i
< 1, then season is smaller than typical
If s
i
> 1, then season is larger than typical
When an actual observation is divided by its corresponding seasonal factor, it is said to be
deseasonalized. (i.e. the seasonal component has been removed.) This allows us to
make meaningful comparisons across time periods.
Let c = the number of periods in a cycle (12 if months of year, 7 if days of week, )
The relevant formulas for this method follow.
E
i
= o( y
i
/ S
i-c
)+(1-o) (E
i-1
+ T
i-1
)
T
i
= |(E
i
L
i-1
) + (1-|)T
i-1

S
i
= (y
i
/ L
i
) + (1-) s
i-c

yhat
i+1
= (E
i
+ T
i
) s
i+1-c

where is another smoothing constant between 0 and 1
This means that Holt Winters' smoothing is similar to exponential smoothing if | and =
0. It will be similar to double exponential smoothing if = 0.
To start this method, we need L
1
, T
1
, and a seasonal factor for each period in the cycle.
An easy way for developing initial estimates of the seasonal factors is to collect c
observations and let:
s
i
= y
i
/ [ (1/c)(y
1
+ y
2
+ y
c
)]
Then assume that E
c
= y
c
/s
c
and that T
c
= 0.
So the steps of the process can be summarized as:
i. forecast for period i
ii. collect observation for period i
iii. calculate smoothed average and the error for period i
iv. calculate the trend for period i
v. calculate the seasonal factor for period i
vi. i = i+1, go back to step 1
19

If you want to forecast k periods ahead, use the following logic
yhat
i+k
= (E
i
+ kT
i
) s
i+k-c
Lets start building the model. Goto Worksheet (triple exp solution)

Figure 1.18
Here we have the monthly sales of a company selling fishing rods. The sales peak during
fishing season in spring and summer. Therefore we know that it is a yearly season where
c = 12. The first step is to transform or deseasonalize the first year sales data in C4:C15.
So we enter the formula =C4/((1/12)*SUM($C$4:$C$15)) in cell F4 and fill down to
F15. This is actually how s
i
= y
i
/ [ (1/c)(y
1
+ y
2
+ y
c
)] is entered as Excel Function.
Secondly we enter the formula, E
i
= o( y
i
/ S
i-c
)+(1-o) (E
i-1
+ T
i-1
) in cell D16 which
is =$J$16*C16/F4+(1-$J$16)*(D15+E15) and fill down till D39.
20

After that we enter this formula, T
i
= |(E
i
L
i-1
) + (1-|)T
i-1
in cell E16 which is
=$J$18*(D16-D15)+(1-$J$18)*E15 and fill down till E39.
The next step is entering the formula S
i
= (y
i
/ L
i
) + (1-) s
i-c
in cell F16 which is
=$J$20*C16/D16+(1-$J$20)*F4 and fill down till F39.
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

21

Figure 1.20
MSE at cell I41 is 49.17 now compare to 221.91 initially. o = 0.46 in cell J16, | = 0.05 in
J18 and = 1 in cell J20. (see Figure 1.21 below)
Figure 1.21
22

We have made a big improvement after optimizing with Excel Solver.
If at month 36 we sell 80 fishing rods, how do you forecast the production at month 37?
To forecast simply fill down the formula in cell G39 to cell G40 or enter the formula
=(D39+E39)*F28. Here the result is 38.56 rods or round up to 39.
If you want to forecast k periods ahead, use the following logic.
yhat
i+k
= (E
i
+ kT
i
) s
i+k-c

In this example we want to forecast sales at month 44 i.e. 8 months ahead, so we enter
=(D39+8*E39)*F35 as you can see in cell G47. The result is 328.86 or round up to 329
fishing rods. (see Figure 1.21 above)
E
i
= D39
T
i
= E39
s
i+k-c
= F35
k = 8
c = 12
As you can see with this example, the triple exponential smoothing method is a little
more complicated to build and should give us very good result.

Conclusion
A moving average is commonly used with time series data to smooth out short-term
fluctuations and highlight longer-term trends or cycles. The threshold between short-term
and long-term depends on the application, and the parameters of the moving average will
be set accordingly. For example, it is often used in technical analysis of financial data,
like stock prices, returns or trading volumes. It is also used in economics to examine
gross domestic product, employment or other macroeconomic time series.
Exponential smoothing has proven through the years to be very useful in many
forecasting situations. It was first suggested by C.C. Holt in 1957 and was meant to be
used for non-seasonal time series showing no trend. He later offered a procedure (1958)
that does handle trends. Winters(1965) generalized the method to include seasonality,
hence the name "Holt-Winters Method" or Triple Exponential Smoothing.
All these forecasting methods are very basic but very useful. Time series forecasting
methods can be more advanced than those considered in our examples above. These are
23

based on AutoRegressive Integrated Moving Average (ARIMA) models (also known as
Box-Jenkins technique). Essentially these assume that the time series has been generated
by a probability process with future values related to past values, as well as to past
forecast errors. To apply ARIMA models the time series needs to be stationary. A
stationary time series is one whose statistical properties such as mean, variance and
autocorrelation are constant over time. We will learn how to model ARIMA as a
forecasting method in Chapter 2.

24

Chapter 2:

AutoRegressive Integrated Moving Average / Box-Jenkins
technique

Introduction
Box-Jenkins Methodology Or Arima Forecasting Method: Box-Jenkins forecasting
models are based on statistical concepts and principles and are able to model a wide
spectrum of time series behavior. The underlying goal of this self- projecting time series
forecasting method is to find an appropriate formula so that the residuals/errors are as
small as possible and exhibit no pattern. The model- building process involves four steps.
Repeated as necessary, to end up with a specific formula that replicates the patterns in the
series as closely as possible and also produces accurate forecasts. (The term Arima and
Box-Jenkin are use interchangeably)
It has a large class of models to choose from and a systematic approach for identifying
the correct model form. There are both statistical tests for verifying model validity and
statistical measures of forecast uncertainty. In contrast, traditional forecasting models
offer a limited number of models relative to the complex behavior of many time series
with little in the way of guidelines and statistical tests for verifying the validity of the
selected model. (you learn this in Chapter 1)
Basic Model: With a stationary series in place, a basic model can now be identified.
Three basic models exist, AR (autoregressive), MA (moving average) and a combined
ARMA in addition to the previously specified RD (regular differencing) combine to
provide the available tools. When regular differencing is applied together with AR and
MA, they are referred to as ARIMA, with the I indicating "integrated" and referencing
the differencing procedure.
ARIMA models are widely used in predicting stock prices, company sales, sunspot numbers,
housing starts and many other fields. ARIMA models are also univariate, that is, they are
based on a single time series variable. (There are multivariate models which are beyond the
scope of this book and will not be discussed)

ARIMA processes appear, at first sight, to involve only one variable and its own
history. Our intuition tells us that any economic variable is dependent on many
other variables. How then can we account for the relative success of the Box Jenkins
methodology. The use of univariate forecasts may be important for several reasons:

- In some cases we have a choice of modeling, say, the output of a large number of
processes or of aggregate output, leaving the univariate model as the only feasible
approach because of the sheer magnitude of the problem.
25

- It may be difficult to find variables which are related to the variable being
forecast, leaving the univariate model as the only means for forecasting.

- Where multivariate methods are available the univariate method provides a
yardstick against which the more sophisticated methods can be evaluated.

- The presence of large residuals in a univariate model may correspond to ab-
normal eventsstrikes etc.

- The study of univariate models can give useful information about trends long-
term cycles, seasonal effects etc in the data.

- Some form of univariate analysis may be a necessary prerequisite to multivariate
analysis if spurious regressions and related problems are to be avoided.

While univariate models perform well in the short term they are likely to be out-
performed by multivariate methods at longer lead terms if variables related to the
variable being forecast fluctuate in ways which are different to their past behavior.

Box and Jenkins have developed procedures for this multivariate modeling. However, in
practice, even their univariate approach, sometimes, is not as well understood as the
classic regression method. The objective of this book is to describe the basics of
univariate Box- Jenkins models in simple and layman terms.

The Mathematical Model

ARMA models can be described by a series of equations. The equations are somewhat
simpler if the time series is first reduced to zero-mean by subtracting the sample mean.
Therefore, we will work with the mean-adjusted series

y(t) = y(t) - (2.1)

Where y(t) is the original time series, is its sample mean, and y(t) is the mean-adjusted
series. One subset of ARMA models are the so-called autoregressive, or AR models. An
AR model expresses a time series as a linear function of its past values. The order of the
AR model tells how many lagged past values are included. The simplest AR model is the
first-order autoregressive, or AR(1), model

y(t) = a(1)*y(t-1) + e(t) (2.2)

where y(t) is the mean-adjusted series in period t, y(t-1) is the series in the previous
period value, a(t) is the lag-1 autoregressive coefficient, and e(t) is the noise. The noise
also goes by various other names: the error, the random-shock, and the residual. The
residuals e(t) are assumed to be random in time (not autocorrelated), and normally
distributed. We can see that the AR(1) model has the form of a regression model in which
Y
Y
26

y(t)

is regressed on its previous value. In this form, a(t) is analogous to the regression
coefficient, and e(t) to the regression residuals. The name autoregressive refers to the
regression on self (auto).

Higher-order autoregressive models include more lagged y(t) terms as predictors. For
example, the second-order autoregressive model, AR(2), is given by

y(t) = a(1)*y(t-1) + a(2)*y(t-2) (2.3)

where a(1) , a(2)

are the autoregressive coefficients on lags 1 and 2. The p
th
order autoregressive
model, AR(p) includes lagged terms on period t 1 to t-p .

The moving average (MA) model is a form of ARMA model in which the time series is
regarded as a moving average (unevenly weighted) of a random shock series e(t) . The
first-order moving average, or MA(1), model is given by

y(t) = e(t) + c(1)*e(t-1) (2.4)

where e(t)
,
e(t-1) the residuals at period t and t-1, and c(1) is the first-order moving
average coefficient. As with the AR models, higher-order MA models include higher
lagged terms. For example, the second order moving average model, MA(2), is

y(t) = e(t) + c(1)*e(t-1) + c(2)*e(t-2) (2.5)

The letter q is used for the order of the moving average model. The second-order moving
average model is MA(q) with q = 2 .

We have seen that the autoregressive model includes lagged terms on the time series
itself, and that the moving average model includes lagged terms on the noise or residuals.
By including both types of lagged terms, we arrive at what are called autoregressive-
moving-average, or ARMA, models.

The order of the ARMA model is included in parentheses as ARMA(p,q), where p is the
autoregressive order and q the moving-average order. The simplest, and most frequently
used ARMA model is ARMA(1,1) model

y(t) = d + a(1)*y(t-1) + e(t) c(1)*e(t-1) (2.6)

The general autoregressive moving average process with AR order p and MA order q
can be written as

y(t) = d+ a(1)*y(t-1) + a(2)*y(t-2) + + a(p)*y(t-p) e(t) c(1)*e(t-1) c(2)*e(t-2) -- c(p)*e(t-p) (2.7)

The parameter d will be explained later in this Chapter.

27

ARIMA MODELING

The purpose of ARIMA modeling is to establish a relationship between the present value
of a time series and its past values so that forecasts can be made on the basis of the past
values alone.

Stationary Time Series: The first requirement for ARIMA modeling is that the time
series data to be modeled are either stationary or can be transformed into one. We can
define that a stationary time series has a constant mean and has no trend overtime. A plot
of the data is usually enough to see if the data are stationary. In practice, few time series
can meet this condition, but as long as the data can be transformed into a stationary
series, an Arima model can be developed. (Ill explain this concept in details later in this
Chapter)

We emphasise again that, to forecast a time series using this approach to forecasting, we
need to know whether the time series is stationary. If it is not, we need to make it
stationary as otherwise the results will not make much sense. Not only this, in order to
produce accurate and acceptable forecasts, we need to determine the class and the order
of the model, i.e. whether it is an AR, MA or ARMA model and how many AR and MA
coefficients (p and q) are appropriate. The analysis of the autocorrelation and partial
autocorrelation functions provides clues to all of these questions. Both requirements
above will be calculated and implemented in two Excel spreadsheet examples later.

The general steps for ARIMA modeling are shown in the chart below:
28

29

THE MODELING PROCESS

Box-Jenkins or Arima modeling of a stationary time series involves the following four
major steps:

A) Model identification
B) Model estimation
C) Diagnostic Checking
D) Forecasting

The four steps are similar to those required for linear regression except that Step A is a
little more involved. Box-Jenkins uses a statistical procedure to identify a model, which
can be complicated. The other three steps are quite straightforward. Let's first discuss the
mechanics of Step A, model identification, which we would do in great detail. Then we
will use an example to illustrate the whole modeling process.

A) MODEL IDENTIFICATION

ARIMA stands for Autoregressive- Integrated-Moving Average. The letter "I"
(Integrated) indicates that the modeling time series has been transformed into a stationary
time series. ARIMA represents three different types of models: It can be an AR
(autoregressive) model, or a MA (moving average) model, or an ARMA which includes
both AR and MA terms. Notice that we have dropped the "I" from ARIMA for simplicity.

Let's briefly define these three model forms again.

AR Model:
An AR model looks like a linear regression model except that in a regression model the
dependent variable and its independent variables are different, whereas in an AR model
the independent variables are simply the time-lagged values of the dependent variable, so
it is autoregressive. An AR model can include different numbers of autoregressive terms.
If an AR model includes only one autoregressive term. it is an AR ( 1 ) model; we can
also have AR (2), AR (3), etc. An AR model can be linear or nonlinear. Below are a few
examples:

AR(1)

y(t) = d + y(t-1) + e(t) (2.8)

AR(3)

y(t) = d + a(1)*y(t-1) + a(2)*y(t-2) + a(3)*y(t-3) + e(t) (2.9)

I will explain more on the d later.

MA Model:
30

A MA model is a weighted moving average of a fixed number of forecast errors produced
in the past, so it is called moving average. Unlike the traditional moving average, the
weights in a MA are not equal and do not sum up to 1. In a traditional moving average,
the weight assigned to each of the n values to be averaged equals to 1 /n; the n weights
are equal and add up to 1. In a MA, the number of terms for the model and the weight for
each term are statistically determined by the pattern of the data; the weights are not equal
and do not add up to 1. Usually, in a MA the most recent value carries a larger weight
than the more distant values,

For a stationary time series, one may use its mean or the immediate past value as a
forecast for the next future period. Each forecast will produce a forecast error. If the
errors so produced in the past exhibit any pattern, we can develop a MA model. Notice
that these forecast errors are not observed values; they are generated values. All MA
models, such as MA (1), MA (2). MA (3), are nonlinear. Below are a few examples:

MA(1)

y(t) = e(t) + c(1)*e(t-1) (2.10)

MA(2)

y(t) = e(t) + c(1)*e(t-1) + c(2)*e(t-2) (2.11)

ARMA Model:
An ARMA model requires both AR and MA terms. Given a stationary time series, we
must first identify an appropriate model form. Is it an AR, or a MA or an ARMA? How
many terms do we need in the identified model? To answer these questions we can use
two methods

1) We can use a subjective way by calculating the autocorrelation function and
the partial autocorrelation function of the series.

2) Or use objective methods of identifying the best ARMA model for the data at
hand. (Automated Arima)

Method 1

i) What are Autocorrelation Function (ACF) and Partial Autocorrelation Function
(PACF)?

Understanding the ACF and PACF is very important in order to use method (1) to
identify which model to use. Without going into the mathematics, ACF values fall
between -1 and +1 calculated from the time series at different lags to measure the
significance of correlations between the present observation and the past observations,
and to determine how far back in time (i.e., of how many time-lags) are they correlated.
31

PACF values are the coefficients of a linear regression of the time series using its lagged
values as independent variables. When the regression includes only one independent
variable of one-period lag. the coefficient of the independent variable is called first order
partial autocorrelation function; when a second term of two period lag is added to the
regression, the coefficient of the second term is called the second order partial
autocorrelation function, etc. The values of PACF will also fall between -1 and +1 if the
time series is stationary.

Lets me show you how to calculate the ACF and PACF with an example. Open the
workbook Arima in the Chapter 2 folder. Select Worksheet (acf). This worksheet contains
Dow Jones Industrial Composite Index (DJI) daily closing stock values between 20 July
2009 and 29 September 2009. All in all, this includes 51 daily values for the series.

ACF
Below is the general formula for Autocorrelation (ACF):
(2.12)

Dont be intimidated by this formula. It is easily implemented in a spreadsheet using
Excel function. We can simplify this procedure by using some of the built in Excel
formula. Formula above essentially tells us that the autocorrelation coefficient for some
lag k is calculated as the covariance between the original series and the series removed k
lags, divided by the variance of the original series.

Excel contains both the covariance and variance function, and they are: =VAR(range),
and, =COVAR(range, range). Worksheet (acf) contains details how these two functions
can be used to calculate autocorrelation coefficients.

Figure 2.0
32

From the formula (and we only show the first seven values and calculations) it is clear
that the variance part is easy, i.e. just the range $C$2:$C$52 in our case. The covariance
is just a bit more difficult to calculate. The ranges are:

$C$2:C51, C3:$C$52
$C$2:C50, C4:$C$52
$C$2:C49, C5:$C$52
$C$2:C48, C6:$C$52, etc.

This means that if we copy the cells downwards, C51 will become C52, then C53 etc. To
avoid this problem, we can copy the formula down the column, but we need to manually
change C51 onwards in a descending sequence. There you go. The ACF values are
calculated in column D.

PACF
The PACF plot is a plot of the partial correlation coefficients between the series and lags
of itself. A partial autocorrelation is the amount of correlation between a variable and a
lag of itself that is not explained by correlations at all lower-order-lags. The
autocorrelation of a time series Y at lag 1 is the coefficient of correlation between Y(t)
and Y(t-1), which is presumably also the correlation between Y(t-1) and Y(t-2). But if
Y(t) is correlated with Y(t-1), and Y(t-1) is equally correlated with Y(t-2), then we should
also expect to find correlation between Y(t) and Y(t-2). (In fact, the amount of correlation
we should expect at lag 2 is precisely the square of the lag-1 correlation.) Thus, the
correlation at lag 1 "propagates" to lag 2 and presumably to higher-order lags. The partial
autocorrelation at lag 2 is therefore the difference between the actual correlation at lag 2
and the expected correlation due to the propagation of correlation at lag 1.

Select Worksheet (pacf). This show how PACF is implemented and calculated in Excel.
The PACF values are specify in Column C. The partial autocorrelation coefficients are
defined as the last coefficient of a partial autoregression equation of order k. This is the
general formula.
(2.13)
Where t > 1, is the autocorrelation, t is the PACF
The formula above is implemented on cell E4, F5, G6, H7, I8 and so on. (see Fig 2.1
below)

Figure 2.1
33

We can see that the PACF calculation is a bit difficult and complex. Fortunately for you
Ive written a macro to simplify the calculation. To use this macro, you need to load
nn_Solve into your Excel. I will show you the steps on how to do it with an example
later. (see Appendix A on how to load nn_Solve)

ii) How do we use the pair of ACF and PACF functions to identify an appropriate model?

A plot of the pair will provide us with a good indication of what type of model we want
to entertain. The plot of a pair of ACF and PACF is called a correlogram. Figure 2.2
shows three pairs of theoretical ACF and PACF correlograms.

Figure 2.2

In modeling, if the actual correlogram looks like one of these three theoretical
correlograms, in which the ACF diminishes quickly and the PACF has only one large
spike, we will choose an AR (1) model for the data. The " 1" in parenthesis indicates that
34

the AR model needs only one autoregressive term, and the model is an AR of order 1.
Notice that the ACF patterns in 2a and 3a are the same, but the large PACF spike in 2b
occurs at lag 1, whereas in 3b, it occurs at lag 4. Although both correlograms suggest an
AR (1) model for the data, the 2a and 2b pattern indicates that the one autoregressive
term in the model is of lag 1; but the 3a and 3b pattern indicates that the one
autoregressive term in the model is of lag 4.

Suppose that in Figure 2.2, ACF and PACF exchange their patterns, that is, the patterns
of PACF look like those of the ACF and the patterns of ACF look like the PACF having
only one large spike, then we will choose a MA (I) model. Suppose again that the PACF
in each pair looks the same as the ACF, and then we will try an ARMA(1, 1).

So far we have described the simplest AR, MA, and ARMA models. Models of higher
order can be so identified, of course, with different patterns of correlograms.

Although the above catalogue is not exhaustive, it gives us a reasonable idea of what to
expect when deciding about the most basic models. Unfortunately the above behavioural
catalogue of autocorrelation and partial autocorrelation functions is only theoretical. In
practise, the actual autocorrelations and partial autocorrelations only vaguely follow these
patterns, which is what makes this subjective approach to forecasting very difficult. In
addition to this, the real-life time series can be treated as just a sample of the underlying
process. Therefore, the autocorrelations and partial autocorrelations that are calculated are
just estimates of the actual values, subject to sampling error.

The autocorrelations and partial autocorrelations also play a prominent role in deciding
whether a time series is stationary, to what class of models it belongs and how many
coefficients it is characterised by. The question that is still open is how to calculate the
coefficients a and c that constitute a particular model.

Before we proceed with how to estimate a and c, we shall return to the question of
differencing and stationarity as promised earlier. In general we must be cautious where
differencing is concerned, which will influence the class of the model. It would be wrong
to assume that when unsure as to whether the series is nonstationary, it should simply be
differenced. Over differencing can lead us to believe that the time series belongs to a
completely different class, which is just one of the problems.

Rules For Differencing
How do we, then, know if we have exaggerated and overdifferenced the series? One of the
basic rules is: if the first autocorrelation of the differenced series is negative and more than
0.5, the series has probably been overdifferenced. Another basic rule is: if the variance for
the higher level of differencing increases, we should return to the previous level of
differencing. One of the rules of thumb is that the level of differencing corresponds to the
degree of a polynomial trend that can be used to fit the actual time series.

The whole notion of differencing is closely related to the concept of so-called unit roots.
Unit root means that an AR(1) or a MA(1) coefficient is equal to one (unity). For higher
order models, this means that the sum of all coefficients is equal to one. If this happens we
35

have a problem. If an AR(1) model has a unit root, then this AR coefficient should be
eliminated and the level of differencing should be increased. For higher AR(p) models, the
number of AR coefficients has to be reduced and the level of differencing increased. For
MA models showing unit roots, an MA coefficient should also be removed, but the level of
differencing has to be decreased. Sometimes we do not catch unit roots early enough,
and produce forecasts, which turn out to be very erratic. This is also a consequence of unit
roots, which means that a reduction in AR or MA coefficients is necessary.

Another question we need to answer is: what is the meaning of d, how do we calculate it
and when do we include it in a model?

Essentially, d in ARMA models plays the same role as the intercept in linear regression.
Our model here is called an ARMA model with a level, where d represents this initial level
of the model (an intercept). Sometimes it is also referred to as the trend parameter, or a
constant.

If we want to calculate this trend parameter, we need to start with the formula for the
expected value of an AR process, i.e. the mean value. The mean of any AR(p) process is
calculated as:

Z = d / (1- a(1) - - a(p)) (2.17)

Which, for AR(2) yields:

Z = d / (1- a(1) a(2)) (2.18)

From this formula, the level d (or the trend component) for the AR(2) process is calculated
as:
d = z * (1 a(1) a(2) (2.19)

In general, the level for any AR(p) process is calculated as:

d = z *(1- ) (2.20)

Now we know what it is and how to calculate it, the open part of the question is still: when
do we include it in our model?

The set of rules can be summarised as follows:

- If a time series is non-stationary in its original form and we have had to difference
it to make it stationary, then the constant is usually not needed
- Time series differenced more than twice do not need a constant
- If the original time series is stationary with zero mean, a constant is not needed.
1
( )
p
i
a p
=
36

- If the original series is stationary, but with a significantly large mean (which
effectively means >1), the constant is necessary.
o If the model does not have an AR component (i.e. it is an MA or IMA
model), then the constant is equal to the mean value of the series.
o If the model has an AR component, the constant is calculated as in (2.20)

(I will show you another example where we will calculate the constant d later)

Testing for zero mean to indicate stationarity
What happens if one level of differencing is not enough and the second level is too
much? This sometimes happens in practise and a time series appears to be stationary, yet
its mean value is not zero, despite the stationarity requirement that it should. If this
happens, we have to ensure that the mean is at least close to zero. The easiest way to do
this is to calculate the mean (the average) of the differenced series w
t
, and subtract it
from every observation: (Goto Worksheet (Daily Sales) ). The data from A2:A101 is the
daily sales from a sport shop in thousand.
implemented in column B

Once we have transformed the differenced time series in such a way, we can calculate
this transformed series mean value, in cell E3 and check whether it is zero. How do
we check whether the mean is zero or close to zero? First we need to estimate this
transformed series standard error. You will remember that SE is a ratio between the
standard deviation and the square root of the number of observations:
(2.14)
The transformed time series mean, , is considered nonzero if:
(2.15) enter in cell E5

Dont worry about the math symbols above. They can be easily implanted in an Excel
spreadsheet. (see Figure 2.3 below). The result is non-zero (see cell E5) and from the
chart we can see that the time series appears nonstationary i.e. its trending up. So we
need to pre-process the time series which is differencing. A one lag differencing i.e. y(t)
= y(t) y(t-1). is applied. The values in C2:C100 are the 1 lag differencing value.

Figure 2.3
x
x
w
z
z
37

There is another common approach to transformations, which avoids differencing. In
finance, for example, we are often more interested in returns, i.e. if we sell shares today
(y
t
), how much we have earned when compared with when we bought them (y
t-1
).
Mathematically this is simply: (y
t
-y
t-1
)/y
t-1
. Even if the shares values are jumping wildly,
the series of such calculated returns will usually be stationary. The above mathematical
expression is known to be approximately equal to log(y
t
)-log(y
t-1
), which is often used for
calculating returns. This expression can also be used to transform a time series into a
stationary form. Some stationary series are not strictly stationary and although they have a
constant mean, their variance is not constant (remember the idea of homoscedasticity?).
The log transformation suggested here is known to reduce heteroscedasticity.

After a stationary series is in place, a basic model can now be identified. Three basic
models exist, AR (autoregressive), MA (moving average) and a combined ARMA in
addition to the previously specified RD (regular differencing) combine to provide the
available tools. When regular differencing is applied together with AR and MA, they are
referred to as ARIMA, with the I indicating "integrated" and referencing the differencing
procedure.

Bear in mind that we are using method (1) to identify the model. So far I have 3
components that are important for us to understand in order to identify the model.

- The ACF and PACF
- Data stationary
- Differencing

Let's use a spreadsheet example to show how to calculate the ACF and PACF first and
then to demonstrate what we have just discussed i.e using ACF and PACF to determine
the p and q parameters as in ARMA(p,q)

We copy the one lag differenced values in C2:C100 and paste them to range C2:C100 in
Worksheet (Daily Sales(1)). From the chart below it now look stationary and random.
And the test also indicates that the time series has zero mean in cell L4

Figure 2.4
38

Now we need to calculate the ACF and PACF. Although, Ive show you how to calculate
them manually (see worksheet(acf) and worksheet (pacf), it is still very troublesome,
especially when you calculate PACF. Fortunately, you can use the nn_Solve addin
written by me to calculate the ACF and PACF automatically. Load nn_Solve to your
Excel. (see Appendix on how to load nn_Solve)

1) Select Arima on the nn_Solve menu (see Figure 2.4a)

Figure 2.4a
Enter the reference that you want to calculate in the Data Range. In our case, we enter
C2:C100. (see Figure 2.4b below). The data range cannot start with row 1 like C1,
A1, B1 and so on. Else nn_Solve will give you an error. Always enter the data that
you want to calculate starting from row 2 like C2, A2, B2 and so on in a spreadsheet.

Figure 2.4b

2) Then click on the Calculate button. The ACF, PACF and the Standard Error will
be calculated. (see Figure 2.4c)

39

Figure 2.4c
Build the charts below using the data calculated. (see Fig. 2.5 and Fig 2.6).The
autocorrelation and the partial autocorrelation functions for the differenced sales revenue
data are given in Fig. 2.5 and Fig 2.6

Figure 2.5

Figure 2.6

The partial autocorrelation function in Fig. 2.6 shows two coefficients as significantly non-
zero, implying that this is an ARMA(p,q) model. The autocorrelation function confirms this
assumption as it shows the pattern usually associated with an ARMA(p,q) model. Given that
40

we had to difference the original time series, the model we will use, therefore, is
ARIMA(2,1,1) model or ARMA(2,1).

B) MODEL ESTIMATION

The equation for this model is:

y(t) = d + a(1)*y(t-1) + a(2)*y(t-2) e(t) - c(1)*e(t-1) (2.16)

Lets implement this formula in a spreadsheet to optimise the coefficients, fit the model and
produce the forecasts.. Open worksheet (Daily Sales(2)). The 1 lag difference sales figures
are enter in column A. In column B is the residual. Column C is the full formula. Press Ctrl
+ ~ to view the formula in your Excel sheet. (see Figure 2.7 below)

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
41

Figure 2.8

The initial values of a(1), a(2) and c(1) are set in cells E2, E3 and E4. Cells E5 and E6
contain the time series mean and standard deviation. Since we have applied differencing
to the time series, the value d is not needed and enters as 0 in cell E8. I will show you
another example where we will calculate the d later when we use method (2).

Our sales revenue data set was originally nonstationary and had to be differenced before
modelling could be applied. This is the reason for omitting the constant d in the first place.
So we set d as 0 in this example. (See Fig 2.8 above)

From the formula 2.16 we can easily extract e(t), which is:

e(t) = y(t) d + a(1)*y(t-1) + a(2)*y(t-2) c(1)*e(t-1)

The above formula implies that to calculate e(1). But we need to know e(0), which we do
not. The convention is to assign zeros to all the unknown values of e(0). In Fig. 2.7
above, we can see zero in cell B2 and B3, which is the first cell needed to perform this
calculation. Since the model is a ARMA(2,1), we also assign a 0 to B3.

Now we have all the errors e(t) , given just the initial values of a(1), a(2) and c(1), we can
calculate the so-called conditional sum of squares of residual (SSE), which is conditional
on the values of a(1) and c(1). The formula for SSE is:
42

SSE(a,c) =

Cell E10 gives us the value of SSE= 377.07 initially, which was obtained using Excel
function =SUMSQ(B2:B100).This cell is instrumental for estimating the optimum value
of a(1), a(2) and c(1), which will hopefully lead towards best possible forecast. To
achieve this, we will use Excel Solver. Our objective is to minimise SSE (i.e. the value in
cell E10), by changing values of E2:E4, i.e. the values of a(1), a(2) and c(1). As before,
we need to define the admissible region which will guarantee that our model is stationary
and invertible. For ARIMA(2,1,1) processes, this is: -1<a(1)<1 and 1<c(1)<1, or,
|a(1)|<1 and |c(1)|<1. Cells E12 to E15 define these conditions.

Before we show how to use Solver, we need to understand one more point about AR(p)
coefficients a(1), a(2), etc. A process that is generated using these coefficients has to be
stationary. In other words, certain values of a(1), a(2), etc., will not necessarily generate a
stationary process. To satisfy this strict condition of stationarity, we need to define the
admissible region for these coefficients. In case of AR(1), this admissible region is defined
as: 1<a(1)<1 (or, |a(1)|<1). In case of AR(2), this admissible region is defined by three
conditions:

a(2)

+ a(10)

< 1, a(2)

a(1)

< 1 and -1 < a(2)

< 1 (or, |a(2)| < 1) and -1 < c(1)

< 1

We can see that our initial estimates of a(1) , a(2) , c(1) in Fig. 2.8 satisfy all these
stationarity conditions. These parameters are entered in cell E12 to E15. One last thing
before I show how to use Solver to calculate the coefficients.

Now we understand modelling (at least for this class of models), we must establish
whether the estimated values of the model coefficients are truly the best ones available.
Traditionally this question involves very complex and complicated calculations, which
ensure that the maximum likelihood estimators are selected. Fortunately with help from
Excel Solver, many of these operations are not necessary. Lets do it

Our objective is to minimize the SSE value in cell E10.

Cell E10 gives us the value of SSE = 377.07 initially, which was obtained using Excel
function =SUMSQ(B2:B100).

This cell, together with cells E12 to E15 is instrumental for estimating the optimum value
of a(1), a(2) and c(1), which will hopefully lead towards best possible forecast. To
achieve this, we will use Excel Solver. Our objective is to minimise SSE (i.e. the value in
cell E10), by changing values of E2:E4, i.e. the values of a(1), a(2) and c(1). As before,
we need to define the admissible region which will guarantee that our model is stationary
and invertible. For ARIMA(2,1,1) processes, this is: : -1<a(1)<1 and 1<c(1)<1, or,
|a(1)|<1 and |c(1)|<1. Cells E12 to E15 define these conditions.

1
( )
n
t
e t
=
43

Upon invoking Solver from the Tools menu, a dialogue box appears as in Fig 2.9 below

Figure 2.9

Next we will enter all the parameters in this dialogue box. It will look like this after all
the parameters are entered. (see Fig 2.10)

Figure 2.10

a) Set Target Cell : E10 (the SSE)
b) By Changing Cells : E2:E4 (a(1), a(2), c(1))
c) The constraint as shown in cell E12:E15

44

Figure 2.11

Before Optimization

d) Click the Solve button. Solver will start to optimize.

Figure 2.12
e) And Keep the Solver solution

45

Figure 2.13
After optimization

The solution is instantly found and the values appear as you can see in Fig 2.13 above

As we can see a(1) is now -0.537871274, a(2) is 0.058098633 and c(1) becomes
0.614100745, which gives a much lower value of SSE = 137.09, compared with the
previous value of 377.07

As we have calculated the e(t), implicitly in column B, if we really wanted it, we can
explicitly calculate the values of the fitted time series, i.e. the ex-post forecasts y(t) in
accordance with this model. We use the formula:

y(t) = d + (-0.537871274*y(t-1)) + 0.058098633*y(t-2) - 0.614100745*e(t-1) (2.18)

instead of

y(t) = d + a(1)*y(t-1) + a(2)*y(t-2) e(t) - c(1)*e(t-1)

You can see that I have drop the e(t) from our formula as we have calculated them in
column B to derive the values in Column C. Column C in Fig. 2.15 shows the values for
y(t) and Fig. 2.14 shows the formula used to produce Fig. 2.15

46

Figure 2.14

Figure 2.15
47

Figure 2.16

Figure 2.17

How closely the fitted values are matching the original time series can be seen in Fig.2.16
above. Forecasting errors from the column B are displayed in Fig. 2.17 above and they
seem to be randomly distributed, as expected. Before we are ready to forecast, we need to
do a diagnostic checking first.

C) DIAGNOSTIC CHECKING:

How do we know that we have produced a reasonable model and that our model indeed
reflects the actual time series? This is a part of the process that Box and Jenkins refer to as
diagnostic checking. I will use two methods to conduct the diagnostic.
Actual Vs Predicted
-6
-4
-2
0
2
4
6
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
y(t)
y(t) = d + a(1)y(t-1) + a(2)y(t-2) e(t) - c(1)e(t-1)
48

As we expect the forecasting errors to be completely random, the first step is to plot them,
as we did in Fig 2.17 above for example. One of the requirements is that the residual mean
should be zero, or close to zero. To establish that this is the case, we need to estimate the
standard error of the mean error. This is calculated as:

(2.19)

(2.20) in cell E18

Where o
e
is the residual standard deviation, is the mean error, n is the number of errors
and is the standard error of the mean error. If the residual mean is greater than
1.96 standard errors, then we can say that it is significantly non-zero:

(2.21) in cell E20

We can take an example from Column B for which the errors e(t) are calculated and shown
in . How to estimate the standard residual error SE
e
(standard error) is shown below in Fig
2.18 and the formula are given in Fig. 2.19 below

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
n
) e (e
n
1 t
2
t
e
=

= o
n
SE
e
e
=
e
e
SE
e
e
SE 96 . 1 e >
49

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

50

Cell E20 contains a brief IF statement evaluating whether the calculated mean value from
E17 is greater than the standard error times 1.96. In our model we have zero mean which
pass the test.

Another test that is quite popular is the Durbin-Watson test, which is use in the context of
checking the validity of ARIMA models.
The DurbinWatson statistic is a test statistic used to detect the presence of
autocorrelation in the residuals from a regression analysis. It is named after James Durbin
and Geoffrey Watson. If e
t
is the residual associated with the observation at time t, then
the test statistic is
(2.22)
Since w in cell E26 is approximately equal to 2(1-r), where r is the sample
autocorrelation of the residuals,

w = 2 indicates no autocorrelation. The value of w always
lies between 0 and 4. If the DurbinWatson statistic is substantially less than 2, there is
evidence of positive serial correlation. As a rough rule of thumb, if DurbinWatson is
less than 1.0, there may be cause for alarm. Small values of w indicate successive error
terms are, on average, close in value to one another, or positively correlated. If w > 2
successive error terms are, on average, much different in value to one another, i.e.,
negatively correlated. In regressions, this can imply an underestimation of the level of
statistical significance.
Figure 2.20
In our model we have 1.90449 in cell E26 which is very close to 2 which indicates no
autocorrelation. See Fig 2.20 above. We can now proceed with the forecast.

D) FORECAST

Now we are ready to produce real forecasts, i.e. those that go into the future. The
equation can be applied one step ahead to get estimate y(t) from observed y(t-1). A k-
step-ahead prediction can also be made by recursive application of equation. In recursive
application, the observed y at time 1 is used to generate the estimated y at time 2. That
estimate is then substituted as y(t-1) to get the estimated y at time 3, and so on. The k-
step-ahead predictions eventually converge to zero as the prediction horizon, k, increases.
Goto cell A101:A105.

51

We will forecast as per the formula below: Arima(2,1,1) or ARMA(2,1). The formula is

y(t) = -0.537871274*y(t-1) + 0.058098633*y(t-2) - 0.614100745*e(t-1)

Figure 2.21

Figure 2.22

Fig. 2.21 shows the spreadsheet containing basic numbers and Fig 2.22 shows all the
calculations.

Figure 2.23

As we have already explained, once we have run out of actual values, the actual values of
y(t) are replaced by their fitted values (starting from C102). This inevitably degrades
forecasts, and we explained how different models behave. As we can see, our forecast for
Predicted Values
-4
-3
-2
-1
0
1
2
3
4
5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Actual
Predicted
52

cell C102 and C103 in Fig.2.21 is very good (as we know the actual values, we put them
in cells A101:A105). Unfortunately our forecast for cell C104 begins to be significantly
different from the known actual value in cell A104. This implies that for many time series
the Box-Jenkins method is a very good fit, but only for short-term forecasts.

To summarise, in this section we not only showed the whole process of identifying
models, fitting them and forecasting, but we also presented a much quicker way of doing
it. We linked the values of ARMA coefficients directly with the residual sum of squares,
which became a target value in the Solver, and which in one single step produced optimal
values for these coefficients.

Automatic Box Jenkins/ARIMA: Method 2

The subjective procedure outlined above requires considerable intervention from the
statistician/economist completing the forecast. Various attempts have been made to
automate the forecasts. The simplest of these fits a selection of models to the data,
decides which is the best and then if the best is good enough uses that. Otherwise the
forecast is referred back for standard analysis by the statistician/economist. The
selection will be based on a criterion such as the AIC (Akaikes Information Criterion)
and the Bayesian (Schwarz) criterion (BIC or SIC)

Before I proceed to implement the automated model in a worksheet, there are 2 important
Excel functions that I want to explain.

i) SumProduct ()
ii) Offset ()

i) SumProduct ()
In Excel, the SumProduct function multiplies the corresponding items in the arrays and
returns the sum of the results.
The syntax for the SumProduct function is:
SumProduct( array1, array2, ... array_n )
array1, array2, ... array_n are the ranges of cells or arrays that you wish to multiply. All
arrays must have the same number of rows and columns. You must enter at least 2 arrays
and you can have up to 30 arrays.
Note: If all arrays provided as parameters do not have the same number of rows and
columns, the SumProduct function will return the #VALUE! error.
53

If there are non-numeric values in the arrays, these values are treated as 0's by the
SumProduct function.
Let's take a look at an example:
=SumProduct({1,2;3,4}, {5,6;7,8})
The above example would return 70. The SumProduct calculates these arrays as follows:
=(1*5) + (2*6) + (3*7) + (4*8)
You could also reference ranges in Excel.

Based on the Excel spreadsheet above, you could enter the following formula:
=SumProduct(A1:B2, D1:E2)
This would also return the value 70. Another example:

=SumProduct(A1:A4, C1:C4).
This will be (2*1) + (3*2) + (4*3) + (5*4) = 40
54

ii) Offset ()
The OFFSET( ) function returns a cell or range of cells that is a specified number of rows
and/or columns from the reference cell. In this Tutorial we will explain the most common
offset() applications and mistakes that are made using this function in Microsoft Excel.
The syntax for OFFSET () is
OFFSET (cell reference, rows, columns, [ height ], [ width ] )
Components in square brackets can be omitted from the formula.
How does the Excel function OFFSET work?
The OFFSET( ) function returns a cell or range of cells that is a specified number of rows
and/or columns from the reference cell. For specific descriptions of each component,
please see the Help file in Excel.
If either of the rows, columns, height or width components is left blank, Excel
will assume its value to be zero. For example, if the formula is written as OFFSET(C38, ,
1, , ) Excel will interpret this as OFFSET(C38, 0, 1, 0, 0). This can also be written as
OFFSET(C38, , 1) since height and width can be omitted.
Note that if height and width are included in the formula, they cannot be equal zero
or a #REF! error will result.
Examples below illustrate the function. Given the following set of numbers
OFFSET Example 1
OFSET(D10, 1, 2) will give the value in F11 or 7, ie, Excel returns the value in the cell 1
row below and 2 columns to the right of D10

OFFSET Example 2
OFFSET(G12, -2, -2) will give the value in E10 or 2, ie, Excel returns the value in the
cell 2 rows above and 2 columns to the left of G12
55

OFFSET Example 3
OFFSET(F12, , , -2, -3) will return the 2 row by 3 column range D11:F12. Note that the
reference cell F12 is included in this range

OFFSET Example 4
OFFSET(D10, 1, 1, 2, 3) will return the range E11:G12, ie, Excel first calculates
OFFSET(D10, 1, 1) which is E11 (1 row below and 1 column to the right of reference
cell D10), then applies the formula OFFSET(E11, , , 2, 3)

Common problems and mistakes with the OFFSET function
When tracing OFFSET( ) functions, only the reference cell is returned. For example,
when tracing the precedent of OFFSET(D10, 1, 1, 2, 3) the returned cell is D10 and not
E11:G12.
Excel excludes the reference cell when calculating the rows and columns
components, but includes the reference cell when calculating the height and width
components.
This can be confusing, and requires extreme care.
OFFSET() is a complex concept to grasp which reduces user confidence in the model
since it is not easily understood.
56

Combining OFFSET() with Other Functions
Since OFFSET( ) returns a cell or a range of cells, it can be easily combined with other
functions such as SUM( ), SUMPRODUCT( ), MIN( ), MAX( ), etc.
For example, SUM( OFFSET( )) calculates the sum of the cell or range of cells returned
by the OFFSET( ) function. Extending from Example 4 above, SUM (OFFSET (D10, 1,
1, 2, 3)) is equivalent to writing SUM(E11 : G12) (as OFFSET (D10, 1, 1, 2, 3) returns
the range E11 : G12) which equals 54 = 6 + 7 + 8 + 10 + 11 + 12. Similarly, AVERAGE
(OFFSET (D10, 1, 1, 2, 3)) is equivalent to AVERAGE (E11 : G12).
As explained in Method 1, Arima modeling involves 4 major steps.
A) MODEL IDENTIFICATION
As mentioned earlier, there is a clear need for automatic, objective methods of identifying
the best ARMA model for the data at hand. Objective methods become particularly
crucial when trained experts in model building are not available. Furthermore, even for
experts, objective methods provide a very useful additional tool, since the correlogram
and partial correlogram do not always point clearly to single best model. The two most
widely used criteria are the Akaike information criterion (AIC) and the Bayesian
(Schwarz) criterion (BIC or SIC)

Dont worry about the formula above. They are easily implemented in an Excel
worksheet. Lets me show you how to build this automated model identification with a
spreadsheet example. Open Worksheet (Work(2)). The data is from the daily electricity
production in a developing country: million kilowatts per day are enter in cell A2:A501.

57

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
58

Figure 2.24

The coefficients are as follow:

a(1) = L11, a(2) = L10, a(3) = L9
c(1) = M11, c(2) = M10

Or if it is a ARMA(2,1) then the cell in L1 will show a 2 and M1 will be a 1. The
corresponding coefficients are cells L10:L11 for the AR and cell M11 for the MA. (see
the blue numbers in Fig 2.25 below)

Figure 2.25
The INT function in Excel is used to remove all decimal places leaving only the whole
number. Removing decimal places, or the fractional part of a number is necessary to use
Excel Solver for our modeling.

Cell L12 is related to L1 as L1 is a whole number or integer of L12. We need to enter the
function Int() in cell L1 as Excel Solver will return an error if L1 is not an Integer. The
same goes for M1 and M12. They are related for the same reason.

As promised earlier, Ill include the calculation of d in this example. The formula for d is
enter in cell I5. (You can refer on how d is derived by looking at page 35 above.)

59

For easier understanding, the general equation above is broken into 3 parts.

i) formula d is enter in cell I5
ii) a(1)*y(t-1) + a(2)*y(t-2) + + a(p)*y(t-p) enter in column B
iii) and e(t) c(1)*e(t-1) c(2)*e(t-2) -- c(p)*e(t-p) enter in column C

i) Read page 35 above to understand how d is calculated.

The formula for d in I5 is:
=I2*(1-(SUM(OFFSET(L12,-1,0,):OFFSET(L12,-L1,0))))

For a ARMA(2,1), this mean I2 * (1-(SUM(L10:L11)).
For a ARMA(3,2) then it is I2 * (1-(SUM(L9:L11)).

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
60

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
61

X

xc(1)*e(t-1) c(2)*e(t-2) => M11*D3 M10*D2

If q = 3 then the formula is (calculation start from C5)

c(1)*e(t-1) c(2)*e(t-2) c(3)*e(t-3) => M11*D4 M10*D3 M9 D2

etc

As for the residuals or errors e(t), these are enter in column D. (see Fig 2.29 below)

Figure 2.29

And lastly the full formula are enter in column E

- d = I5
- Column B = a(1)*y(t-1) + a(2)*y(t-2) + + a(p)*y(t-p)
- Column C = e(t) c(1)*e(t-1) c(2)*e(t-2) -- c(p)*e(t-p)
- Column D = e(t)

Thus the full formula in column E. (see Fig 2.29 above)

Before we invoke Excel Solver to solve the parameters p, q and their coefficients, let me
explain to you the AIC and BIC as we are using an objective method for model identification
in the next section.

B) MODEL ESTIMATION

Because of the highly subjective nature of the Box-Jenkins methodology, time series
analysts have sought alternative objective methods for identifying ARMA models.
Penalty function statistics, such as Akaike Information Criterion [AIC] or Final
Prediction Error [FPE] Criterion (Akaike, 1974), Schwarz Criterion [SC] or Bayesian
Information Criterion [BIC] (Schwarz, 1978) have been used to assist time series analysts in
reconciling the need to minimize errors with the conflicting desire for model parsimony. These
statistics all take the form minimizing the sum of the residual sum of squares plus a penalty
term which incorporates the number of estimated parameter coefficients to factor in
model parsimony..
62

Akaike Information Criterion (AIC)

Bayesian Information Criterion (BIC)

Assuming there is a true ARMA model for the time series, the BIC and HQC have the
best theoretical properties. The BIC is strongly consistent whereas AIC will usually
result in an overparameterised model; that is a model with too many AR or MA terms
(Mills 1993, p.29). Indeed, it is easy to verify that for n greater than seven the BIC
imposes a greater penalty for additional parameters than does the AIC.

Thus, in practice, using the objective model selection criteria involves estimating a
range of models and the one with the lowest information criterion is selected.
These 2 formulas are entered in cells I11 for the AIC and I12 for the BIC. (See Fig 2.30 below)

Figure 2.30

We also need to define the admissible region which will guarantee that our model is
stationary and invertible. The coefficients of AR models must be within a permissible
region in order to guarantee stationarity and there is also a permissible region for the
coefficients of MA models which guarantees invertibility. Every MA model is stationary
63

by definition, but is invertible only if certain conditions are satisfied. By the way, AR
models are invertible for all values of coefficients, but only stationary if coefficients are
in a particular admissible region. In Fig 2.30 above the admissible region ensuring
stationarity is given in cell I20 and the admissible region ensuring invertibility is given in
cell I21. As we have a generalize model for automated Arima, the formula to ensure
stationarity and invertibility are
-1 1 and -1 1 .

Its now time to use Excel Solver for our Automated Arima(p,q) model

Open the Worksheet(Work(3)). Worksheet(Work(3) is just a copy of Worksheet(Work(2)).
To use the Solver, click on the Tools heading on the menu bar and select the Solver . . .
item. (see Figure 2.31)

Figure 2.31

Figure 2.32
s p
s s q
s
64

If Solver is not listed (see Figure 2.31) then you must manually include it in the
algorithms that Excel has available. To do this, select Tools from the menu bar and
choose the "Add-Ins . . ." item. In the Add-Ins dialog box, scroll down and click on the
Solver Add-In so that the box is checked as shown Figure 2.32 above:

After selecting the Solver Add-In and clicking on the OK button, Excel takes a moment
to call in the Solver file and adds it to the Tools menu.

If you cannot find the Solver Add-In, try using the Macs Find File or Find in Windows
to locate the file. Search for solver. Note the location of the file, return to the Add-Ins
dialog box (by executing Tools: Add-Ins), click on Select or Browse, and open the
Solver Add-In file.

What if you still cannot find it? Then it is likely your installation of Excel failed to
include the Solver Add-In. Run your Excel or Office Setup again from the original CD-
ROM and install the Solver Add-In. You should now be able to use the Solver by
clicking on the Tools heading on the menu bar and selecting the Solver item.

Although Solver is proprietary, you can download a trial version from Frontline Systems,
the makers of Solver, at www.frontsys.com.

After executing Tools: Solver . . . , you will be presented with the Solver Parameters
dialog box below:

Figure 2.33

Let us review each part of this dialog box, one at a time.

Set Target Cell is where you indicate the objective function (or goal) to be optimized.
This cell must contain a formula that depends on one or more other cells (including at
65

least one changing cell). You can either type in the cell address or click on the desired
cell. Here we enter cell A11.

In our Arima model, the objective function is to minimize the AIC in cell I11. See Figure
2.34 below.

Equal to: gives you the option of treating the Target Cell in three alternative ways. Max
(the default) tells Excel to maximize the Target Cell and Min, to minimize it, whereas
Value is used if you want to reach a certain particular value of the Target Cell by
choosing a particular value of the endogenous variable.

Here, we select Min as we want to minimize the AIC. (You may also try I12 the BIC)

For starting value, I use p and q = 5. The coefficients = 0.1 (see Fig 2.34 below).

Figure 2.34

By Changing Cells permits you to indicate which cells are the adjustable cells (i.e.,
endogenous variables). As in the Set Target Cell box, you may either type in a cell
address or click on a cell in the spreadsheet. Excel handles multivariable optimization
problems by allowing you to include additional cells in the By Changing Cells box. Each
noncontiguous choice variable is separated by a comma. If you use the mouse technique
(clicking on the cells), the comma separation is automatic.
66

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
67

Figure 2.37

As you can see, a series of choices are included in the Solver Options dialog box that
direct Solvers search for the optimum solution and for how long it will search. These
options may be changed if Solver is having difficulty finding the optimal solution.
Lowering the Precision, Tolerance, and Convergence values slows down the algorithm
but may enable Solver to find a solution.

For a Arima model, you can set :

i) Max Time: 1000 seconds
ii) Iterations: 1000
iii) Precision: 0.000001
iv) Tolerance: 5%
v) Convergence: 0.0001

Select Conjugate as the Search method. This prove to be very effective in minimizing
the AIC.

The Load and Save Model buttons enable you to recall and keep a complicated set of
constraints or choices so that you do not have to reenter them every time.

Clicking OK to return to the Solver Parameters dialog box.

Solve is obviously the button you click to get Excel's Solver to find a solution. This is the
last thing you do in the Solver Parameters dialog box. So, click Solve to start training.

68

Figure 2.38

Figure 2.39

When Solver start optimizing, you will see the Trial Solution at the bottom left of your
spreadsheet. See Figure 2.39 above.

Figure 2.40

A message will appears after Solver has converged (see Figure 2.40)..In this case, Excel
reports that Solver has converged to the current solution. All constraints are satisfied.
This is good news!

Sometime, the solution is not satisfactory and Solver unable to find the solution at one go.
For example it may failed the stationary test as indicated in cell I9 i.e not a zero mean. If
this is the case then you, change the starting parameters for p, q and the coefficients and
run Solver again. Follow the step discussed above. From my experience, usually you will
need to run Solver a few times before Solver arrive at a satisfactory solution.

69

Bad news is a message like, Solver could not find a solution. If this happens, you must
diagnose, debug, and otherwise think about what went wrong and how it could be fixed.
The two quickest fixes are to try different initial p, q parameters and their coefficients.

From the Solver Results dialog box, you elect whether to have Excel write the solution it
has found into the Changing Cells (i.e., Keep Solver Solution) or whether to leave the
spreadsheet alone and NOT write the value of the solution into the Changing Cells (i.e.,
Restore Original Values). When Excel reports a successful run, you would usually want it
to Keep the Solver Solution. On the right-hand side of the Solver Results dialog box,
Excel presents a series of reports. The Answer, Sensitivity, and Limits reports are
additional sheets inserted into the current workbook. They contain diagnostic and other
information and should be selected if Solver is having trouble finding a solution.

Figure 2.41

My first run of Excel Solver come to the above solution. AIC = -6.236111282 in cell I11.
As indicated in the Figure 2.41 above, we have a ARMA(1,1) model. The coefficients are
in cells L11 and M11. It has passed all the tests as you can see cells I9 and H18 in Figure
2.41. (Note: Depending on the data you have, sometimes you need to run Solver a few
times before you come to a satisfactory solution).

70

C) DIAGNOSTIC CHECKING:

How do we know that we have produced a reasonable model and that our model indeed
reflects the actual time series? This is a part of the process that Box and Jenkins refer to as
diagnostic checking. I will use two methods to conduct the diagnostic. As we expect the
forecasting errors to be completely random, the first step is to plot them, as we did in Fig.
2.42 below for example. This residuals chart indicates randomness. But we want to make
sure so we need to do the calculation.

Figure 2.42

One of the requirements is that the residual mean should be zero, or close to zero. To
establish that this is the case, we need to estimate the standard error of the mean error. This
is calculated as:

in cell I7

Where o
e
is the residual standard deviation, is the mean error, n is the number of errors
and is the standard error of the mean error. If the residual mean is greater than
1.96 standard errors, then we can say that it is significantly non-zero:

in cell I9

Residuals/Errors
-4.00
-3.00
-2.00
-1.00
0.00
1.00
2.00
3.00
4.00
1 24 47 70 93 116 139 162 185 208 231 254 277 300 323 346 369 392 415 438 461 484
Series1
n
) e (e
n
1 t
2
t
e
=

= o
n
SE
e
e
=
e
e
SE
e
e
SE 96 . 1 e >
71

How to estimate the standard residual error SE
e
(standard error) is shown below in Fig 2.43
and the formulas are given in Fig. 2.44 below

Figure 2.43

72

Figure 2.44

Cell I9 contains a brief IF statement evaluating whether the calculated mean value from I6
is greater than the standard error times 1.96. In our model we have zero mean which pass
the test.

Another test that is quite popular is the Durbin-Watson test, which is use in the context of
checking the validity of ARIMA models. The DurbinWatson statistic is a test statistic
used to detect the presence of autocorrelation in the residuals from a regression analysis. It
is named after James Durbin and Geoffrey Watson.
If e
t
is the residual associated with the observation at time t, then the test statistic is

H16 contain the upper part of the above formula and H17 contain the lower part. Since w
in cell H18 is approximately equal to 2(1-r), where r is the sample autocorrelation of the
residuals,

w = 2 indicates no autocorrelation. The value of w always lies between 0 and 4.
If the DurbinWatson statistic is substantially less than 2, there is evidence of positive
serial correlation. As a rough rule of thumb, if DurbinWatson is less than 1.0, there may
be cause for alarm. Small values of w indicate successive error terms are, on average,
73

close in value to one another, or positively correlated. If w > 2 successive error terms are,
on average, much different in value to one another, i.e., negatively correlated. In
regressions, this can imply an underestimation of the level of statistical significance.
In our model we have 1.96812 in cell H18 which is very close to 2 which indicates no
autocorrelation. See Fig 2.43 above. We can now proceed with the forecast.

D) FORECAST

Now we are ready to produce real forecasts, i.e. those that go into the future. The
equation can be applied one step ahead to get estimate y(t) from observed y(t-1). A k-
step-ahead prediction can also be made by recursive application of equation. In recursive
application, the observed y at time 1 is used to generate the estimated y at time 2. That
estimate is then substituted as y(t-1) to get the estimated y at time 3, and so on. The k-
step-ahead predictions eventually converge to zero as the prediction horizon, k, increases.
On worksheet Work(3) goto cell A502:A507.

We will forecast as per the formula below: Arima(1,0,1) or ARMA(1,1)

We use the formula:

y(t) =1.30335 + 0.73951*y(t-1) - 0.32419*e(t-1)

Figure 2.45

Figure 2.46
Fig. 2.45 shows the spreadsheet containing the forecasted values numbers and Fig 2.46
shows all the calculations and formulas.
74

Figure 2.47

As we have already explained, once we have run out of actual values, the actual values of
y(t) are replaced by their fitted values (starting from E503). This inevitably degrades
forecasts, and we explained how different models behave. As we can see, our forecast for
cell E502 and E505 in Fig 2.45 is very good (as we know the actual values, we put them
in cells A502:A507 to compare). Unfortunately our forecast for cell E506 begins to be
significantly different from the known actual value in cell A506. This implies that for
many time series the Arima method is a very good fit, but only for short-term forecasts.

You may not have such an ideal model. It takes me about 10 run of Excel Solver on the
model before I came to this result. Change the p, q and their coefficients starting values
and then run Solver until you come to a satisfactory solution. It takes a little bit of testing
and running.

Another way you can make use of the model is use Excel Solver to optimize the
coefficients only. You enter the p and q manually and use Solver to optimize the
coefficient. Let me give you an example. Open Worksheet (Work(4)). Enter 2 in both
cells L12 and M12. (see Fig 2.48 below). So we have use an ARMA(2,2) model. Invoke
Excel Solver and enter the parameters as shown in Fig 2.49 below.

Actual vs Predicted
4.20
4.40
4.60
4.80
5.00
5.20
5.40
5.60
1 2 3 4 5 6
Actual
Predicted
75

Figure 2.48

Figure 2.49

76

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
77

Seasonal Arima Modeling (SARIMA):

We can use ARIMA models for seasonal time series forecasting. The underlying
principles are identical to the ones for non-seasonal time series, described above. These
seasonal time series show seasonal trends with periodicity s

Seasonal series repeat themselves after a certain number of months, usually after twelve
months, or every four months (quarterly seasonality). They can be either stationary or non-
stationary. For the non-stationary seasonal time series, they need to be differenced.
Unfortunately, ordinary differencing is not good enough for such cases. Seasonal
differencing is what is needed.

For example,

- Monthly data has 12 observations per year
- Quarterly data has 4 observations per year
- Daily data has 5 or 7 (or some other number) of observations per week.

A SARIMA process has four components:

- auto-regressive (AR),
- moving-average (MA),
- one-step differencing,
- and seasonal differencing.

For a time series with a 12-month pattern, seasonal differencing is executed as follows:

The differencing formula requires that in a seasonal time series, we need to find
differences between two comparable months, rather than between two successive months
as it makes more sense. In this example, 12 is the number of months. If we assign letter s
for seasonality, then the seasonal differencing is in general described as:

w(t) = y(t) y(t-s)

Like ordinary differencing, sometimes a second level of differencing is needed. This is
done as:

Vw(t)

= w(t) w(t-s)

If we substitute Vw(t)

= w(t) w(t-s) w(t) = y(t) y(t-s) ,we get:

w(t) = [y(t) y(t-s)] [y(t-1) y(t-s-1)] = y(t) y(t-1) y(t-s) + y(t-s-1)

Which for example for s =12 gives:

Vw(t)

= y(1) y(t-1) y(t-12) + y(t-13)
78

The above formula shows that: y(t) = y(t-1) + y(t-12) y(t-13), i.e. in this case the current
observation is equal to the previous observation, plus the one twelve periods ago, less the
one that preceded it! Sounds odd but if we rewrite it a little differently it will make a lot of
sense:

y(t) - y(t-12) = y(t-1) y(t-13)

Thus we are saying that this periods seasonal differences are the same as the seasonal
differences observed in the previous period, which are more logical.

We can make an interesting digression here and ask ourselves what the next periods
seasonal differences are going to be like. It is reasonable to assume that they will be
something like: y
t+1
- y
t-11
= y
t
- y
t-12
, which is very interesting because we can see above,
that y
t
- y
t-12
= y
t-1
- y
t-13
. Essentially we are saying that y
t+1
- y
t-11
= y
t-1
- y
t-13
. Does this
mean that y
t+1
- y
t-11
= y
t
- y
t-12
? Yes, this means that the forecasting origin will determine
all the future seasonal differences.

Let us return to modelling seasonal time series. The above explanations implied that in
order to fit a seasonal time series with an ARIMA model, it is not enough to just have a
model of order (p,d,q). We also need a seasonal order (P,D,Q), which will be combined
with these non-seasonal coefficients (p,d,q). The general formula is

ARIMA(p,d,q)(P,D,Q)
S

How do we combine the two? We can use for example, an SARIMA(1,1,1)(1,1,1)
4
,

i.e. a
model with s = 4. This model is described as:

(1-|
1
B)(1-u
1
B
4
) (1-B) (1-B
4
)y
t
= (1-u
1
B)(1-O
1
B
4
)e
t

Where | and u are the ordinary ARMA coefficients, u and O are the seasonal ARMA
coefficients and B is the backshift operator. If we unravel the above equation, we get:

y
t
= (1+|
1
)y
t-1
- |
1
y
t-2
+ (1+|
1
)y
t-4
(1+|
1
+u
1
+|
1
u
1
)y
t-5
+ (|
1
+|
1
u
1
)y
t-6
- u
1
y
t-8
+
+ (u
1
+|
1
u
1
)y
t-9
- |
1
u
1
y
t-10
+ e
t
- u
1
e
t-1
- O
1
e
t-4
+u
1
O
1
e
t-5

As we can see, it is quite a long and messy equation. We should use abbreviated notation
instead as it makes a lot of sense to use it. So, a seasonal ARIMA(p,d,q)(P,D,Q)
S
model
can be written in a short general form as:

(1-|
1
B)(1-u
1
B
S
)y
t
= (1-u
1
B)(1-O
1
B
S
)e
t

which is much more elegant. An ARIMA(2,1,0)(0,1,0)
12
model, for example, is therefore
written as:

(1-|
1
B-|
2
B
2
)(1-u
1
B
12
)(1-B)y
t
= e
t
79

Where, (1-|
1
B-|
2
B
2
) represents a non-seasonal AR(2) part of the model, (1-u
1
B
12
)
represents the seasonal AR(1) part and (1-B) are non-seasonal differences.

The seasonal parameters and coefficients are ,, P, D, Q and,, p, d, q are for the
non seasonal time series. s denotes the seasonality.

Following the three steps of the Box and Jenkins methodology (Box and Jenkins 1976),
identification, estimation and diagnostics checking the SARIMA models are fitted to
stationary or weakly stationary time-series data. The estimation of the AR (p, P) and MA
(q, Q) parameters for fitting a SARIMA model is about the same when you model an
Arima model.

For example a SARIMA (1,0,0)(0,1,1)12 model, in short form is

(1-|
1
B)(1-B
12
)y
t
= (1-O
1
B
12
)e
t

which leads to

y
t
= |
1
y
t-1
+ y
t-12
- |
1
y
t-12
+ e
t
- O
1
e
t-12

For example a SARIMA (0,1,1)(0,1,1)4 model, in short form is

(1-B) (1-B
4
)y
t
= (1-u
1
B) (1-O
1
B
4
)e
t

which leads to

y
t
= y
t-1
+ y
t-4
- y
t -5
+ u
1
e
t-1
- O
1
e
t-4
+ u
1
O
1
e
t-5

What should we expect from the autocorrelation and partial autocorrelation functions for
these models? In many ways they are identical, in terms of inference, as non-seasonal
models. An ARIMA(0,0,0)(1,0,0)
12
model, for example, will have one significant partial
autocorrelation at lag 12 and the autocorrelations will decay exponentially for all seasonal
lags, i.e. 12, 24, 36, etc. An ARIMA(0,0,0)(0,0,1)
12
model, on the other hand, will have
one significant autocorrelation at lag 12 and the partial autocorrelations will decay
exponentially for all seasonal lags.

The principles of parameter estimation for seasonal models are the same as for the non-
seasonal models, although the equations for SARIMA can be messier.

Please be aware that it is impractical and unnecessary to do seasonal differencing of the
series twice. It is good practice not to difference the time series more than twice, regardless
of what kind of differencing is used, i.e. use a maximum of one seasonal and one ordinary
differencing or, at most, do ordinary differencing twice. One of the most popular and
frequently used seasonal models in practice is an ARIMA(0,1,1)(0,1,1)
S
. Most time series
can be fitted with an ARIMA(0,1,1)(0,1,1)
S
, so do not exaggerate.
80

Conclusions:

ARIMA model offers a good technique for predicting the magnitude of any variable. Its
strength lies in the fact that the method is suitable for any time series with any pattern of
change and it does not require the forecaster to choose a priori the value of any
parameter.

ARIMA models also provide useful tools for interested parties to be used as a point of
reference to the performance of other forecasting models like neural network, kernel
regression and so on. However, please bear in mind that the forecasting inaccuracy
increased the farther away the forecast is from the used data, which is consistent with the
expectation of ARIMA models. It takes lots of practice and experimenting. Hopefully with all
the examples presented in this Chapter can speed up and shorten your learning curve.

81

Chapter 3

Monte Carlo Simulation With MS Excel

Introduction
Monte Carlo simulation is a widely used method for solving complex problems using
computer algorithms to simulate the variables in the problem. Typically an algorithm is
developed to "model" the problem, and then the algorithm is run many times (from a few
hundred up to millions) in order to develop a statistical data set to study how the model
behaves. This statistical data simulated can be represented as probability distributions (or
histograms) or converted to error bars, reliability predictions, tolerance zones, and
confidence intervals.
The Monte Carlo method is also one of many methods for analyzing uncertainty
propagation, where the goal is to determine how random variation, lack of knowledge,
or error affects the sensitivity, performance, or reliability of the system that is being
modeled. Monte Carlo simulation is categorized as a sampling method because the
inputs are randomly generated from probability distributions to simulate the process of
sampling from an actual population. So, we try to choose a distribution for the inputs that
most closely matches data we already have, or best represents our current state of
knowledge.
For example, consider the basic coin toss where we have two possible outcomes (heads or
tails), each with a 50% probability. In a million coin tosses, roughly half will be "heads"
and half will be "tails". No complex math is required to know this. A simple Monte
Carlo simulation would prove the same result. If you were to develop a spreadsheet with
a random number generator resulting in 0 for "heads" and 1 for "tails" (which we will do
later in this chapter), then make the spreadsheet to recalculate a million times, each time
recording the results in a database, you could then run a report on the database which will
show that very close to 50% of the recalculations resulted in 0 or "heads" and the other
50% in 1 or "tails".
In a more complicated coin tossing simulation, suppose you want to know the likelihood
of getting "heads" 7 times out of 10 coin tosses. Again here, statistical mathematical
equations provide an accurate answer without the need for Monte Carlo. A Monte Carlo
spreadsheet can simulate a series of ten coin tosses as easy as a single coin toss, and can
keep a record of how many times 7 "heads" were returned from 10 tosses after running a
few thousand iterations of the model. The resulting data will have an answer very close
to the mathematical statistical probability of getting 7 "heads".
82

Although we can represent and solve this problem using mathematical equations, a
problem with this degree of complexity becomes more efficient to solve using a Monte
Carlo simulation.
How the Monte Carlo Algorithm works?
Monte Carlo Algorithm works based on the Law of Large Numbers. It says that if you
generate large number of samples, eventually you will get the approximate desire
distribution.
- To use Monte Carlo Simulation is quite simple and straightforward as long as the
convergence can be guaranteed by the theory.
- Monte Carlo Simulation may provide statistical sampling for numerical
experiment using computer
- For optimization problems, Monte Carlo Simulation often can reach global
optimum and overcome local extreme
- Monte Carlo Simulation provides approximate solution to many mathematical
problems
- The method can be used for both stochastic (involve probability) and
deterministic (without probability).
That's essentially all there is to it. The more complex the problem, the better it will be to
use Monte Carlo simulations over mathematical solving. Most of the times the problem
may involve some guessing at how a particular variable behaves (is it a normal
distribution curve, poison, or a linear?), but by using Excel spreadsheets to build your
models, it's easy to change the assumptions behind each variable to study the sensitivity
the results on each of your assumptions. I will use 3 examples to show you how to
implement a Monte Carlo simulation on a spreadsheet.
Before we develop the Excel spreadsheet example, let me explain to you the RAND()
function which is the most essential building block of Monte Carlo models in Excel. The
function is Excel's native random number generator. This function returns an evenly
distributed random real number greater than or equal to 0 and less than 1. A new random
real number is returned every time the worksheet is calculated. To use the RAND()
function, simply enter "=RAND()" in the spreadsheet cell. Each time the spreadsheet is
updated or recalculated a new random number will be generated.
Note: The results in this book and the example spreadsheets can be different due to
Excel recalculation.
Example 1: Coin tossing
Developing Monte Carlo models require one to translate a real-world problem into Excel
equations. This is a skill you can develop over time. The examples later in this chapter
will give you a good start.
83

As a first example, look at the case of a single coin toss. In the real world we think in
terms of "heads" and "tails", but for data analysis it will be more efficient to represent
these outcomes to 0 and 1. Next you have to figure out how to convert the output
generated by the RAND() function into the output of a coin toss -- i.e., take evenly
distributed randomly generated numbers greater than or equal to 0 and less than 1, and
translate them into two single outcomes - 0 or 1 - with a 50% probability of each.
As your model variables increase in complexity, however, it's important to master the art
of RAND() manipulation. For the coin toss, a simple approach would be
"=ROUND(RAND(),0)". Another approach that works equally well is
"=INT(RAND()*2)".
Building a worksheet-based simple coin toss Monte Carlo simulation
As I have covered the basic of Monte Carlo concepts, it's time to build the first Monte
Carlo simulation worksheet.
We will simulate a single coin toss. In order to determine the probability of the coin
landing heads-up or tails-up, we will repeat the simple coin toss many times, then
calculate the percentage of those tosses that yield heads. Open the workbook (Monte
Carlo) in folder Chapter 3and goto Worksheet (Coin 1)
First we enter the formula simulating a single toss, "=INT(RAND()*2)", as illustrated in
cell A4 below.

Figure 3.1
Next, we copy cell A4 one thousand times, filling cells A4 through A1003. To do this,
make sure cell A4 is highlighted, then just fill down until the selected region reaches cell
A1003.
Now we'll calculate the results. In cell E2, enter the formula
"=AVERAGE(A4:A1003)". Format the cell to show results as a percentage. The result
should be somewhere near 50%. (see Figure 3.2)
84

Figure 3.2
So we have completed a Monte Carlo simulation in its simplest form. The "Average"
function tells you that approximately 50% of the coin tosses will result in "heads".
How many iterations are enough?
In this example, you probably noticed the answer didn't come out to exactly 50%, even
though 50% is the statistically correct answer.
Every time you recalculate the spreadsheet by pressing the F9 key, you will see the
average change with each recalculation. You may see instances where the average is as
low as 46% or as high as 54%. But by increasing the number of iterations in your model,
you will increase the accuracy of your results.
To prove this, we create another 5000 instances of coin tossing. Make a new column B of
simulated coin tosses, by pasting & copying "=INT(RAND()*2)" into 5000 cells. As
shown below, the column with 5000 iterations returns an answer much closer to 50% than
the column with only 1,000 iterations. (see Figure 3.3)

Figure 3.3
One method to determine whether you should increase the number of iterations is to
recalculate the spreadsheet several times, and observe how much variance there is in the
"Average". If you see more variance in the results than you are satisfied with, increase
85

the number of iterations until this spot-check method gives answers within an acceptably
narrow range.
Thus, the number of iterations depends on the complexity of the variables and the
required accuracy of the result.
Building a worksheet-based multiple coin toss Monte Carlo simulation
We will take the simple coin toss simulation a step further.
Suppose we want to know:

1. What is the probability of getting "heads" on seven out of ten tosses, and
2. What is the probability of getting "heads" on at least seven out of ten tosses?
Like the case of the single coin toss above, we can actually solve the problem
mathematically rather than through simulation. In this ten coin tosses there are exactly
2^10, or 1024, possible outcomes of heads-tails combinations. Of these, exactly 120
have seven "heads", so the probability of getting exactly seven "heads" in ten coin tosses
is 120/1024, or 11.7%. Furthermore, 176 of the 1024 combinations have 7, 8, 9, or 10
heads, so the probability of getting heads on at least seven tosses is 176/1024, or 17.2%.
For this example, we use the same formula to simulate a single toss that we used in the
previous example. Goto Worksheet (Coin 2). We enter the formula "=INT(RAND()*2)"
into cells A10 through J10, as shown in Figure 3.4 below.

Figure 3.4
Then we sum the results of columns A through J by entering "=SUM(A10:J10)" in cell
L10 as shown above. This represents the number of "heads" that came up in ten random
tosses of a coin.
86

Now make 5,000 copies of row 10 in the rows below it. We fill down from A10:L10 to
A5009:L5009. Thus we have a Monte Carlo worksheet-based simulation representing
5,000 iterations of tossing a coin ten times.
To study the results, we make use of the COUNT() and COUNTIF() functions. As
indicated below in Fig 3.5, in cell D3 we enter the formula "=COUNT(L10:L5009)" to
return the exact number of iterations in the model. We enter the formula
"=COUNTIF(L10:L5009,7)" in cell D4 to count the number of iterations which returned
exactly seven heads. Cell E4 returns the percent of all iterations that resulted in seven
heads "=D4/D3", and in this particular instance the outcome was 11.88% - which is quite
close to the expected statistical outcome of 11.7%.
The formula "=COUNTIF(L15:L5014,">6")" in cell D5 counts the number of iterations
which returned AT LEAST seven heads. E5 shows that result as a percent of all
iterations, and again our outcome of 17.88% is also quite close to the statistical
probability of 17.2%. (see Figure 3.5 below)

Figure 3.5
With this simple Monte Carlo example, we have now the basic knowledge to build a
more complicated model that deals with sales forecasting.

X
X
X
87

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
88

Figure 3.6

Figure 3.7
In Column I is the Sales Volume with maximum value 80000 and minimum value 50000.
Column J is Unit Cost with maximum value 8 and minimum value 5. And Column K is
the Selling Price with maximum value 12 and minimum value 8. (see Figure 3.7)
Column M shows the profit. There are 1000 rows which mean we are simulating 1000
sales scenarios, each row using a different combination of values for the sales volumes,
sales price and unit costs. (You can simulate more than 1000 rows if you want)
We enter =INT($D$4 + RAND()*($E$4-$D$4)) in cell I2 which is
the minimum value of Sales Volume in D4
plus a random value (derive with the Excel formula RAND()
times
(maximum value of Sales Volume in cell E4
minus the minimum value of Sales Volume in D4)
Since we are simulating 1000 scenarios, simply copy the formula down 1000 rows until
cell I1001, making sure that you use relative references in the formula (the $ signs).
89

We also enter the same basic formula for the Unit Cost and Selling Price. Activate cell J2
and K2 to view the formulas. The general formula to generate random number with an
upper and lower bound is
r = Int(minimum + Rand() * (maximum minimum))
We enter =I2*(K2-J2)-$B$3 in cell M2 and fill down to M1002 as the profit formula.
Here there are 1000 possible profit values. (Note: you can simulate more than 1000 profit
values). Because we have used the volatile RAND() formula, to re-run the simulation all
we have to do is recalculate the worksheet (press F9 is the shortcut).
After that, a histogram is build using the simulated profit values to further analyze the
data.
How to create a Histogram in Excel
First, the Data Analysis ToolPak must be installed. To do this, pull down the Tools menu
from the Excel menu, and choose Add-Ins. You will see the Analysis ToolPak inside the
Add-Ins dialog box.

To start, you need to have a column of numbers in the spreadsheet that you wish to create
the histogram from, AND you need to have a column of intervals or "Bin" to be the upper
boundary category labels on the X-axis of the histogram.
Cell O7:O57 is the bin values. Its contain 50 ranges/bins between the maximum profit
100000 and the minimum profit 55000. (see part of the formula below)

Figure 3.7a
Pull Down the Tools Menu and Choose Data Analysis, and then choose Histogram and
click OK. Enter the Input Range as M2:M1001 and enter the Bin Range as O7:O57.
Choose whether you want the output in a new worksheet ply, or in a defined output range
on the same spreadsheet. (see Figure 3.8 below)
90

Figure 3.8
If you choose in the example above to have the output range P7:P57, and you clicked
OK, the spreadsheet would look like this:
Figure 3.9
91

The next step is to make a bar chart of the Frequency column. Select Q8:Q58 and click
on the graph Wizard, and choose Column Graph, click on Finish. Delete the Series
legend, right click on the edge of the graph and choose Source Data , and enter the Bin
frequencies (P8:P58) for the X-Axis Category labels. (see Figure 3.9a)

Figure 3.9a
Dress up the graph by right clicking on the edge of the graph and choosing Chart
Options. Enter a complete descriptive title with data source, perhaps data labels, and
axes labels. You may also right click and format the color of the bars and background.
The completed Histogram should look something like this:
92

Figure 3.10 : A Histogram in Excel created using a Bar Chart.
(From a Monte Carlo simulation using n = 1000 points and 50 bins).
After creating the histogram, the next step is to analyze the results visually.
We can glean a lot of information from this histogram:
- It looks like profit will be positive, most of the time.
- The uncertainty is quite large, varying between -48000 to 381000.
- The distribution does not look like a perfect Normal distribution.
- There doesn't appear to be outliers, truncation, multiple modes, etc.
The histogram tells a good story, but in many cases, we want to estimate the probability
of being below or above some value, or between a set of specification limits.
In our Monte Carlo Simulation example, we plotted the results as a histogram in order to
visualize the uncertainty in profit. In order to provide a concise summary of the results, it
is customary to report the mean, median, standard deviation, standard error, and a few
other summary statistics to describe the resulting distribution. The screenshot below
shows these statistics calculated using simple Excel formulas.

NOTE: The results below and in the worksheet can differ from the Histogram in this
book because when Excel recalculate by pressing F9, the Histogram that we created will
not update automatically according with the recalculated data. We need to rebuild the
Histogram again. Basically, I just want to show how to derive the statistics and how to
visually interpret the Histogram.
93

Figure 3.11 : Summary statistics for the sales forecast example.
As you can see in Figure 3.11, we have:
B12 = Sample Size (n): =COUNT(M:M)
B13 = Sample Mean: =AVERAGE(M:M)
B14 = Sample Standard Deviation (o): =STDEV(M:M)
B15 = Maximum: =MAX(M:M)
B16 = Mininum: =MIN(M:M)
B20 = Q(.75): =QUARTILE(M:M,3)
B19 = Q(.25): =QUARTILE(M:M,1)
B17 = Skewness: =SKEW(M:M)
B18 = Kurtosis: =KURT(M:M)
B21 = Median: =MEDIAN(M:M)
Sample Size (n)
The sample size, n, is the number of observations or data points from a single MC
simulation. For this example, we obtained n = 1000 simulated observations. Because the
Monte Carlo method is stochastic, if we repeat the simulation, we will end up calculating
a different set of summary statistics. The larger the sample size, the smaller the difference
will be between the repeated simulations. (See standard error below).
Central Tendency: Mean and Median
The sample mean and median statistics describe the central tendency or "location" of
the distribution. The arithmetic mean is simply the average value of the observations.
The mean is also known as the "First Moment" of the distribution. In relation to physics,
if the probability distribution represented mass, then the mean would be the balancing
point, or the center of mass.
94

If you sort the results from lowest to highest, the median is the "middle" value or the
50th Percentile, meaning that 50% of the results from the simulation are less than the
median. If there is an even number of data points, then the median is the average of the
middle two points.
Spread: Standard Deviation, Range, Quartiles
The standard deviation and range describe the spread of the data or observations. The
standard deviation is calculated using the STDEV function in Excel.
The range is also a helpful statistic, and it is simply the maximum value minus the
minimum value. Extreme values have a large effect on the range, so another measure of
spread is something called the Interquartile Range.
The Interquartile Range represents the central 50% of the data. If you sorted the data
from lowest to highest, and divided the data points into 4 sets, you would have 4
Quartiles:
Q0 is the Minimum value: =QUARTILE(M:M,0) or just =MIN(M:M),
Q1 or Q(0.25) is the First quartile or 25th percentile: =QUARTILE(M:M,1),
Q2 or Q(0.5) is the Median value or 50th percentile: =QUARTILE(M:M,2) or
=MEDIAN(G:G),
Q3 or Q(0.75) is the Third quartile or 75th percentile: =QUARTILE(M:M,3),
Q4 is the Maximum value: =QUARTILE(M:M,4) or just MAX(G:G).

In Excel, the Interquartile Range is calculated as Q3-Q1 or:
=QUARTILE(M:M,3)-QUARTILE(M:M,1)
Shape: Skewness and Kurtosis
Skewness
Skewness describes the asymmetry of the distribution relative to the mean. A positive
skewness indicates that the distribution has a longer right-hand tail (skewed towards more
positive values). A negative skewness indicates that the distribution is skewed to the left.
Kurtosis
Kurtosis describes the peakedness or flatness of a distribution relative to the Normal
distribution. Positive kurtosis indicates a more peaked distribution. Negative kurtosis
indicates a flatter distribution.
Confidence Intervals for the True Population Mean
The sample mean is just an estimate of the true population mean. How accurate is the
estimate? You can see by repeating the simulation (pressing F9 in this Excel example)
that the mean is not the same for each simulation.
95

Standard Error
If you repeated the Monte Carlo simulation and recorded the sample mean each time, the
distribution of the sample mean would end up following a Normal distribution (based
upon the Central Limit Theorem). The standard error is a good estimate of the
standard deviation of this distribution, assuming that the sample is sufficiently large (n
>= 30).
The standard error is calculated using the following formula:

In Excel: =STDEV(M:M)/SQRT(COUNT(M:M))
95% Confidence Interval
The standard error can be used to calculate confidence intervals for the true population
mean. For a 95% 2-sided confidence interval, the Upper Confidence Limit (UCL) and
Lower Confidence Limit (LCL) are calculated as:

To get a 90% or 99% confidence interval, you would change the value 1.96 to 1.645 or
2.575, respectively. The value 1.96 represents the 97.5 percentile of the standard normal
distribution. (You may often see this number rounded to 2). To calculate a different
percentile of the standard normal distribution, you can use the NORMSINV() function in
Excel.
Example: 1.96 = NORMSINV(1-(1-.95)/2)
Note:
Keep in mind that confidence intervals make no sense (except to statisticians), but they
tend to make people feel good. The correct interpretation: "We can be 95% confident
that the true mean of the population falls somewhere between the lower and upper
limits." What population? The population we artificially created! Lest we forget, the
results depend completely on the assumptions that we made in creating the model and
choosing input distributions. "Garbage in ... Garbage out ..." So, I generally just stick to
using the standard error as a measure of the uncertainty in the mean. Since I tend to use
Monte Carlo simulation for prediction purposes, I often don't even worry about the
mean. I am more concerned with the overall uncertainty (i.e. the spread).
96

As a final step in the sales forecast example, we are going to look at how to use the Excel
percentile function and percent rank function to estimate important summary statistics
from our Monte Carlo simulation results.
Percentile and PercentRank Functions
Excel's PERCENTILE() and PercentRank function are useful in many Monte Carlo
models, and is particularly useful in answering our questions in regards to forecasting
profit . For example:
Question 1: What percentage of the results was less than -$22000?
This question is answered using the percent rank function: =PERCENTRANK(array,x,
significant_digits), where the array is the data range and x is $22000.
You can read more about the details of the RANK, PERCENTILE, and
PERCENTRANK functions in the Excel help file (F1).
The Figure 3.14 below shows a screen shot of some examples where the percent rank
function is used to estimate the cumulative probability based upon results of the Monte
Carlo simulation. (Select the cells in the worksheet to see formula entered). So we have
6.87% of the result that are below -22000. And we know that 60.58% of the profit result
is more than 100000.

Figure 3.14 : Calculating probabilities using the Excel percent rank function.
The accuracy of the result will depend upon the number of data points and how far out
on the tails of the distribution you are (and of course on how realistic the model is, how
well the input distributions represent the true uncertainty or variation, and how good the
random number generator is). Recalculating the spreadsheet a few times by pressing F9
will give you an idea of how much the result may vary between each simulation.
Question 2: What are the 95% central interval limits?
Stated another way: What are the 0.005 and 0.95 quantiles?
This is probably one of the most important questions, since the answer provides an
important summary statistic that describes the spread of the data. The central interval is
97

found by calculating the 0.05 and 0.95 quantiles, or Q(alpha/2) and Q(1-alpha/2),
respectively.
Percentile is a measure that locates where a value stands in a data set. The kth percentile
divides the data so that at least p percent are of this value or less and (100-p) percent are
this value or more. If you have a set of data and need to find the value at a certain
percentile, you use the PERCENTILE function in Excel.
The quantiles (or percentiles) are calculated by using the Excel percentile function:
=PERCENTILE(array,p) where the array is the data range (column M) and p is the
cumulative probability (0.05 or 0.95).
The figure below shows a screen shot of examples that use the percentile function in the
Monte Carlo simulation example spreadsheet.

Figure 3.15 : Calculating quantiles using the Excel percentile function.
Note that we are not using the term "confidence interval" to describe this interval. We are
estimating what proportion of the data we expect to be within the given limits based upon
the results of the simulation. We call it a central interval because we are defining the
interval based upon the central proportion of the data.
NOTE: The results above and in the worksheet may differ from the Histogram in this
book because when Excel recalculate by pressing F9, the result that we created will
update accordingly with the recalculated data. Basically, I just want to show how to
derive the statistics and how to visually interpret the Histogram.
A Few Things to Keep in Mind
Beware that defining the uncertainty of an input value by a probability distribution that
does not correspond to the real one and sampling from it will give incorrect results.
In addition, the assumption that the input variables are independent might not be valid.
Misleading results might come from inputs that are mutually exclusive or if significant
correlation is found between two or more input distributions.
Also note that the number of trials should not be too small, as it might not be sufficient to
simulate the model, causing clustering of values to occur.

There you go. This example shows you how to use Monte Carlo simulation for sales
forecast and planning. This example is not comprehensive, and there are many other
factors that affect sales that have not been covered. However, I hope this model has given
you a good introduction to the basics.
98

Example 3: Modeling Stock Prices
Prior to the 1960s, most investors believed that future securities prices could be
predicted (and that great riches were to be had) if only they could discover the secret.
Many investors still believe this today, despite much evidence that suggests that they
would be best served by simply owning the entire market (investing in index funds)
rather than trying to pick individual stocks.
The efficient markets hypothesis (EMH) essentially states that techniques such as
fundamental and technical analysis cannot be used to consistently earn excess profits in
the long run. The EMH began with the observation that changes in securities prices
appear to follow a random walk (technically, geometric Brownian motion with a positive
drift). The random walk hypothesis was first proposed by mathematician Louis Bachelier
in his doctoral thesis in 1900, and then promptly forgotten.
One of the best-known stories regarding the randomness of changes in stock prices is told
by Burton Malkiel in A Random Walk Down Wall Street (a fascinating practitioner-
oriented book now in its ninth edition). In that book he tells of having fooled a friend,
who was a committed technical analyst, by showing him a "stock chart" that was
generated by coin tosses, rather than actual stock prices. Apparently, the friend was quite
interested in learning the name of the stock.
The purpose of this example is not to debate market efficiency, or to even state that the
EMH is correct. Instead, I want to demonstrate how we can simulate stock prices of the
kind that Malkiel discussed.
The Geometric Brownian Motion Process for Stock Prices

We assume that stock prices follow a (continuous time) geometric Brownian motion
process:

dS= Sd t+ oSd
where,

S = the current stock price
= the expected stock return
o = the stock return volatility
dz= (dt)
0.5
, is a standard random variable: < _(0,1)

The discrete time equivalent is:

S S t S t o c A = A + A
99

A = change
At = change in period
Dont worry about the formula and symbols here. They are easily implemented in Excel
which is shown in the Worksheet (Stock Price).
Goto Worksheet (Stock Price). For our simulation, we will use the closing price of
Google from 19 August 2004 to 6 June 2008. These data are entered in column
A52:B1008. First of all, we calculate the percentage of daily return of the closing price.
We enter the formula (today closing price yesterday closing price)/ today closing price
i.e C52 = (B52-B53) and fill down the formula until cell C1007. Column D shows the
percentages of return. (see Figure 3.16)

Figure 3.16
Next we calculate the average and standard deviation of these percentages of daily
returns. These are entered in cells B19 and B20 respectively.

Figure 3.16a
The idea is using Monte Carlo simulation to resample or replicate the closing prices. We
enter $567 the closing price on 6/6/2008 in cell B17. We will use Monte Carlo to
simulate 300 days into the future.
100

Before I go further, let me explain to you, the Excel function =NORMINV()
Excel does not provide a random number generator which directly produces samples
from a general normal distribution. Instead it is necessary to first generate a uniform
random number from the interval zero to one and then use this number to create a sample
from the desired normal distribution. The good news is that Excel makes it very easy to
perform each of these separate tasks.
The typical Excel formula is:
=NORMINV(RAND(), average, standard deviation)

Applied through many iterations, this formula will yield the normal distribution of data
values described by the specified mean and standard deviation.

The RAND function produces the zero-one random number. The NORMINV function
converts this into the sample from the normal distribution. The second argument to the
NORMINV function is the mean of the distribution, in our case we enter the percentage
of average daily return. The third argument is the standard deviation of the normal
distribution, i.e. we enter the percentage of the standard deviation of the daily return.

Starting from range L6:Q6 we enter the closing price on 6/6/2008 at period 0. This is
=$B$17. (see Figure 3.16b below)

Figure 3.16b
In this example, we are simulating 6 different stock price paths. You can simulate more
stock price paths if you want. In cell L7 i.e period 1, we enter the formula (see Figure
3.17)
=+$B$19*L6*($B$21)+$B$20*L6*NORMINV(RAND(),0,1)*($B$21)^0.5+L6.
This is actually how is entered as Excel formula.

= cell B19 => mean/average
times
S = cell L6 => stock price at previous period
times
S S t S t o c A = A + A
101

x
x
x
x
x
x
x
x
x
x
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
102

Figure 3.18
By repeating the path simulation, you can obtain a complete distribution of stock prices at
the end of 300 days. You can see the 6 stock price paths graphically in Fig 3.18 above.

Press F9 to recalculate the worksheet. You can use the simulated stock price paths to test
your stock trading strategy. Things like maximum drawdown, entry and exit strategy,
value at risk calculation etc can be tested against these stock price paths. Thus you can
optimize the best trading method for maximum return.
In practice, it is usually more convenient to buy an add-on for Excel than to do a Monte
Carlo analysis from scratch every time. But not everyone has the money to spend, and
hopefully the skills you have learned from these examples will aid in future data analysis
and modeling.
Conclusion
The Monte Carlo Simulation technique is straightforward and flexible. It cannot wipe out
uncertainty and risk, but it can make them easier to understand by ascribing probabilistic
characteristics to the inputs and outputs of a model. It can be very useful for determining
different risks and factors that affect forecasted variables and, therefore, it can lead to
more accurate predictions.

103

Chapter 4:

K Nearest Neighbors

Introduction

The K Nearest Neighbor or KNN prediction technique is among the oldest techniques
used in data mining. Most people have an intuition that they understand what clustering
is - namely that like records are grouped or clustered together. Nearest neighbor is a
prediction technique that is quite similar to clustering - its essence is that in order to
predict what a prediction value is in one record look for records with similar predictor
values in the historical database and use the prediction value from the record that it
nearest to the unclassified record. KNN is also part of supervised learning that has been
used in many applications in the field of data mining, statistical pattern recognition,
image processing and many others. Some successful applications include recognition of
handwriting, satellite image and EKG pattern. Instead of using sophisticated software or
any programming language, I will build 3 examples using only spreadsheet functions of
Microsoft Excel. These examples include
- using KNN for classification
- using KNN for time series prediction
- Cross validation method
A simple explanation of nearest neighbor
A simple understanding of the nearest neighbor prediction algorithm is that if you look at
the people in your neighborhood (in this case those people that are in fact geographically
near to you). You may notice that, in general, you all have somewhat similar incomes.
Thus if your neighbor has an income greater than $150,000 chances are good that you too
have a high income. Certainly the chances that you have a high income are greater when
all of your neighbors have incomes over $150,000 than if all of your neighbors have
incomes of $25,000. Within your neighborhood there may still be a wide variety of
incomes possible among even your closest neighbors but if you had to predict
someones income based on only knowing their neighbors youre best chance of being
right would be to predict the incomes of the neighbors who live closest to the unknown
person.
The nearest neighbor prediction algorithm works in very much the same way except that
nearness in a database may consist of a variety of factors not just where the person
lives. It may, for instance, be far more important to know which school someone
attended and what degree they attained when predicting income. The better definition of
near might in fact be other people that you graduated from college with rather than the
people that you live next to.
104

Nearest Neighbor techniques are among the easiest to use and understand because they
work in a way similar to the way that people think - by detecting closely matching
examples. They also perform quite well in terms of automation, as many of the
algorithms are robust with respect to dirty data and missing data.
How to use Nearest Neighbor for Prediction
One of the essential elements underlying the concept of nearest neighbor is that one
particular object (whether they be cars, food or customers) can be closer to another object
than can some other third object. It is interesting that most people have an innate sense of
ordering placed on a variety of different objects. Most people would agree that an apple
is closer to an orange than it is to a tomato and that a Toyota Vios is closer to a Honda
Civic than to a Ferrari. This sense of ordering on many different objects helps us place
them in time and space and to make sense of the world. It is what allows us to build
clusters/neighbors - both in databases on computers as well as in our daily lives. This
definition of nearness that seems to be ubiquitous also allows us to make predictions.
The nearest neighbor prediction algorithm simply stated is:
Objects that are near to each other will have similar prediction values as well. Thus if
you know the prediction value of one of the objects you can predict it for its nearest
neighbors.
Where has the nearest neighbor technique been used in business?
One of the classical places that nearest neighbor has been used for prediction has been in
text retrieval. The problem to be solved in text retrieval is one where the end user defines
a document (e.g. Wall Street Journal article, technical conference paper etc.) that is
interesting to them and they solicit the system to find more documents like this one.
Effectively defining a target of: this is the interesting document or this is not
interesting. The prediction problem is that only a very few of the documents in the
database actually have values for this prediction field (namely only the documents that
the reader has had a chance to look at so far). The nearest neighbor technique is used to
find other documents that share important characteristics with those documents that have
been marked as interesting. Thus we use the k-nn method to classify data. This is only
one example classification. K-nn method is applied in many other areas as well like
recognition of handwriting, satellite image, EKG pattern, stock selection, DNA
sequencing and etc. Let us study K-nearest neighbor algorithm for classification in
details.

a) Using KNN for classification (Example 1)
K-nearest neighbor is a supervised learning algorithm where the result of new instance
query is classified based on majority of K-nearest neighbor category. The purpose of this
algorithm is to classify a new object based on attributes and training samples. The
classifiers do not use any model to fit and only based on memory. Given a query point,
we find K number of objects or (training points) closest to the query point. The
105

classification is using majority vote among the classification of the K objects. Any ties
can be broken at random. K Nearest neighbor algorithm used neighborhood classification
as the prediction value of the new query instance.
Here is step by step on how to compute K-nearest neighbors (KNN) algorithm:
i. Determine parameter K = number of nearest neighbors
ii. Calculate the distance between the query-instance and all the training samples
iii. Determine nearest neighbors based on the K-th minimum distance
iv. Gather the category Y of the nearest neighbors
v. Use simple majority of the category of nearest neighbors as the prediction
value of the query instance Or predict the mean, for numeric prediction

i) What value to use for k? Determine parameter K = number of nearest neighbors
It depends on dataset size. Large data set need a higher k, whereas a high k for a small
dataset might cross out of the class boundaries. Calculate accuracy on test set for
increasing value of k, and use a hill climbing algorithm to find the best. Typically use an
odd number to help avoid ties. Selecting the k value is quite intuitive. You can choose
heuristically the optimal k nearest neighbor based on Mean Squared Error done by a cross
validation technique on a test set. (I will show you this technique in example (3))

Below are 2 graphic examples to determine k:

Figure 4.1

K-nearest neighbors of a record x are data points that have the k smallest distance to x.

106

(source: Wikipedia)
Figure 4.2
ii) Calculate the distance between the query-instance and all the training samples
Open the workbook (KNN) in folder Chapter 4 and goto Worksheet (Home).We have
data from a bank record and objective testing with two attributes (income and lots size)
to classify the home ownership of a customer. Our objective is to determine whether a
customer is entitled a lower interest rate for the second home loan. Here are the training
samples

107

Figure 4.3

First we need to scale and standardize the raw data so that they will have same
dimensionality and range. We use the formula

(Raw data Mean of Raw Data) / Standard Deviation of Raw Data

Range E2:F27 is the transform data. (see Figure 4.3 above). Select respective cells to
view formula entered to do the transformation.

Now there is a customer with yearly income of USD60000 and live in a 20000 square
feet apartment. How do we know whether the customer live in his/her own home without
asking him/her. Fortunately, k nearest neighbor (KNN) algorithm can help you to predict
this type of problem.
The data for this example consist of 2 multivariate attributes namely x
i
(income and lots
size) that will be used to classify y (home ownership). The data of KNN can be any
measurement scale from ordinal, nominal, to quantitative scale but for the moment let us
deal with only quantitative x
i
and binary (nominal) y. Later in this section, I will explain
how to deal with other types of measurement scale.
108

Suppose we have the data in Figure 4.3 above. Row 27 is the query instance that we want
to predict i.e whether the customer is an owner or non owner
Because we use only quantitative x
i
, we can use Euclidean distance. The general formula
for Euclidean distance is:

Lets enter the above formula as Excel function. Select cell H2. You will see the formula
=((E2-E$27)^2+(F2-F$27)^2)^0.5 entered. This is how the above formula looks like as
Excel function. Fill down the formula until cell H25. Thus, we have calculated the
Euclidean distance. (see Figure 4.4 below)

Figure 4.4
109

iii. Determine nearest neighbors based on the K-th minimum distance
Now that we have established a measure in which to determine the distance between two
scenarios, we can simply pass through the data set, one scenario at a time, and compare it to the
query scenario. That is to find the K-nearest neighbors. We include a training sample as
nearest neighbors if the distance of this training sample to the query instance is less than
or equal to the K-th smallest distance. In other words, we rank the distance of all training
samples to the query instance and determine the K-th minimum distance.
If the distance of the training sample is below the K-th minimum, then we gather the
category y of this nearest neighbors' training samples. In MS excel, we can use MS Excel
function =SMALL(array, K) to determine the K-th minimum value among the array.
Let me explain the SMALL(array,k) function in details first before we move on.
The syntax for the Small function is:
Small( array, nth_position )
array is a range or array from which you want to return the nth smallest value.
nth_position is the position from the smallest to return.
Let's take a look at an example:

Based on the Excel spreadsheet above:

110

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
111

v. Use simple majority of the category of nearest neighbors as the prediction value of the
query instance
The KNN prediction of the query instance is based on simple majority of the category of
nearest neighbors. In our example, the category is only binary, thus the majority can be
taken as simple as counting the number of 1' and 0 signs. If the number of 1 is greater
than 0, we predict the query instance as 1 and vice versa. If the number of 1 is equal to
0, we can choose arbitrary or determine as one of the 1 or 0.
To calculate the 1 and 0 in our example, we enter the formula =COUNTIF(I2:I25,1)
for 1 and =COUNTIF(I2:I25,0) for 0 in cell I27 and I28 respectively. Here we have
more 1 than 0. To determine this we enter the formula =IF(I27>=I28,1,0) in cell H30.
This simply say to return 1 if value in cell I27 is more than or equal to I28 else return
0.
The predicted result is 1 i.e. the customer own a home if the income is $60000 per
annum and live in 20000 square feet lot.
If your training samples contain y as categorical data, take simple majority among this
data like the example above. If the y is quantitative, take average or any central tendency
or mean value such as median or geometric mean. We will build another example where
y is quantitative below.

b) Using KNN for time series prediction (Example 2)

Using nearest neighbor for stock market data
As with almost all prediction algorithms, nearest neighbor can be used in a variety of
places. Its successful use is mostly dependent on the pre-formatting of the data so that
nearness can be calculated and where individual records can be defined. In the text
retrieval example this was not too difficult - the objects were documents. This is not
always as easy as it is for text retrieval. Consider what it might be like in a time series
problem - say for predicting the stock market. In this case the input data is just a long
series of stock prices over time without any particular record that could be considered to
be an object. The value to be predicted is just the next value of the stock price.
The way that this problem is solved for both nearest neighbor techniques and for some
other types of prediction algorithms is to create training records by taking, for instance,
10 consecutive stock prices and using the first 9 as predictor values and the 10th as the
prediction value. Doing things this way, if you had 100 data points in your time series
you could create 10 different training records.
You could create even more training records than 10 by creating a new record starting at
every data point. For instance, you could take the first 10 data points and create a record.
Then you could take the 10 consecutive data points starting at the second data point, then
112

the 10 consecutive data point starting at the third data point. Even though some of the
data points would overlap from one record to the next the prediction value would always
be different. In our example, Ill only show 5 initial data points as predictor values and
the 6
th
as the prediction value.
Lets look at the example now. Goto Worksheet (Stock Price). Using the same principle
discussed above, we can extend the K-Nearest Neighbor (KNN) algorithm for prediction
(extrapolation) of quantitative data (e.g. time series). In classification, the dependent
variable y is categorical data. In this example, the dependent variable has quantitative
values.
Here is the step by step again on how to compute K-nearest neighbors KNN algorithm for
quantitative data:
X
X
X
X
X
X
X
X
X
X
X
X
X
X
113

Figure 4.6
1. Determine parameter K = number of nearest neighbors
Suppose we use K = 2 in cell C3.
2. Calculate the distance between the query-instance and all the training samples
Coordinate of query instance is 1.5. As we are dealing with one-dimensional distance, we
simply take absolute value from the query instance to value of x.
For instance for x =1, the distance is | 1 1.5 | = 0.5, for x = 1.2 the distance is | 1.2 1.5|
= 0.3 and so on. (see Figure 4.5 above)
3. Rank the distance and determine nearest neighbors based on the K-th minimum
distance
Since we have use K = 2, the 2
nd
rank value is 0.3 in this example. The nearest neighbors
are 1.1, 1.8 and 1.5 (see Figure 4.5 above)
4. Gather the values of y of the nearest neighbors
The values are 1.1, 1.8 and 1.5
5. Use average of nearest neighbors as the prediction value of the query instance
X
X
X
114

gives the lowest MSE from the test set is the optimal k. Lets look at an example on how
to implement the cross validation.

c) Cross Validation Method Using MSE (Example 3)
Open Worksheet Stock Price (2). The time series in B7:B16 is the daily stock price of a
shipping company.
Ive build 4 test set for this example. Test Set 1 use the prices from 2/17/2009 to
2/21/2009 to predict the price at 2/22/2009, Test Set 2 use the prices from 2/18/2009 to
2/22/2009 to predict the price at 2/23/2009 and so on. The same steps and the formulas in
Example 2 are use to build these test set. (see Figure 4.7 below and select respective cell
to view the formula entered)

Figure 4.7
The errors for each test set are entered in range I30:I33. For example the formula =(M12-
B12)^2 is enter in cell I30. This means that the result of prediction in test set 1 in cell
M12 minus the actual result in cell B12. And then we squared the value. We follow the
same reasoning for test set 2, 3 and 4.
115

After that, we obtain the mean squared error in cell I34 by entering the formula
=AVERAGE(I30:I33). This is the MSE we use for cross validation when we use different
k in cell C3.
For k = 1 in cell C3 we have MSE = 0.053438




As you can see from the result above, we obtain the lowest MSE at 0.000977 when k = 3.
This is how you do the cross validation.
Thus, for our prediction, we enter k = 3 in cell C3, and the result is 1.616666667 as you
can see in cell F17. (see Figure 4.8 below)
116

Figure 4.8
Generally, this is how the cross validation method is implemented. You can use more test
set and different partition of the stock prices if you want.
Conclusion
KNN is a very robust and simple method for data classification and prediction. It is very
effective if the training data is large. However, as we can see from the examples above, it
is quite difficult to determine K beforehand. The computation cost is also quite high
because we need to compute distance of each query instance to all training samples.
Nevertheless, KNN is widely deploy in the area of data mining and can perform well in
many situations.

117

Chapter 5:

Building Neural Network Model With MS Excel

Introduction to Neural Network

Everyone try to forecast the future. Bankers need to predict credit worthiness of
customers. Marketing analyst want to predict future sales. Economists want to predict
economic cycles. And everybody wants to know whether the stock market will be up or
down tomorrow. Over the years, many software have been developed for this purpose and
one such software is the neural network based forecasting application. No, neural
network is NOT a medical term. It is actually a branch of artificial intelligence which
gains much prominence since the start of the millennium.
NN or neural network is a computer software (and possibly hardware) that simulates a
simple model of neural cells in humans. The purpose of this simulation is to acquire the
intelligent features of these cells. In this book, when terms like neuron, neural network,
learning, or experience are mentioned, it should be understood that we are using them
only in the context of a NN as computer system. NN have the ability to learn by example,
e.g. a NN can be trained to recognize the image of car by showing it many examples of a
car or to predict future stock prices by feeding it historical stock prices.
We can teach a neural network to perform these particular tasks by using the following
procedure:
I. We present the network with training examples, which consist of a pattern of
activities for the input units together with the desired pattern of activities for the
output units.
II. We determine how closely the actual output of the network matches the desired
output.
III. We change the weight of each connection so that the network produces a better
approximation of the desired output.
I will show you later, on how to integrate the three steps described above with 5 MS
Excel spreadsheet models. With these examples, you can easily understand NN as a non-
linear forecasting tool. NO MORE complex C++ programming and complicated
mathematic formula(s). I have spent much time and effort to simplify how to use NN as a
forecasting tool for you. You only need to know how to use MS Excel, in modeling NN
as a powerful forecasting method. THATS IT!.
Technical Stuff of neural network that you don't really have to know.
Neural networks are very effective when lots of examples must be analyzed, or when a
structure in these data must be analyzed but a single algorithmic solution is impossible to
formulate. When these conditions are present, neural networks are use as computational
tools for examining data and developing models that help to identify interesting patterns
118

or structures in the data. The data used to develop these models is known as training data.
Once a neural network has been trained, and has learned the patterns that exist in that
data, it can be applied to new data thereby achieving a variety of outcomes. Neural
networks can be used to
- learn to predict future events based on the patterns that have been observed in the
historical training data;
- learn to classify unseen data into pre-defined groups based on characteristics
observed in the training data;
- learn to cluster the training data into natural groups based on the similarity of
characteristics in the training data.
We have seen many different neural network models that have been developed over the
last fifty years or so to achieve these tasks of prediction, classification, and clustering. In
this book we will be developing a neural network model that has successfully found
application across a broad range of business areas. We call this model a multilayered
feedforward neural network (MFNN) and is an example of a neural network trained
with supervised learning.
We feed the neural network with the training data that contains complete information
about the characteristics of the data and the observable outcomes in a supervised learning
method. Models can be developed that learn the relationship between these characteristics
(inputs) and outcomes (outputs). For example, we can develop a MFNN to model the
relationship between money spent during last weeks advertising campaign and this
weeks sales figures is a prediction application. Another example of using a MFNN is to
model and classify the relationship between a customers demographic characteristics and
their status as a high-value or low-value customer. For both of these example
applications, the training data must contain numeric information on both the inputs and
the outputs in order for the MFNN to generate a model. The MFNN is then repeatedly
trained with this data until it learns to represent these relationships correctly.
For a given input pattern or data, the network produces an output (or set of outputs), and
this response is compared to the known desired response of each neuron. For
classification problems, the desired response of each neuron will be either zero or one,
while for prediction problems it tends to be continuous valued. Correction and changes
are made to the weights of the network to reduce the errors before the next pattern is
presented. The weights are continually updated in this manner until the total error across
all training patterns is reduced below some pre-defined tolerance level. We call this
learning algorithm as the backpropagation.
Process of a backpropagation
I. Forward pass, where the outputs are calculated and the error at the output units
calculated.

II. Backward pass, the output unit error is used to alter weights on the output units.
119

Then the error at the hidden nodes is calculated (by back-propagating the error at
the output units through the weights), and the weights on the hidden nodes altered
using these values.
The main steps of the back propagation learning algorithm are summarized below:
Step 1: Input training data.
Step 2: Hidden nodes calculate their outputs.
Step 3: Output nodes calculate their outputs on the basis of Step 2.
Step 4: Calculate the differences between the results of Step 3 and targets.
Step 5: Apply the first part of the training rule using the results of Step 4.
Step 6: For each hidden node, n, calculate d(n). (derivative)
Step 7: Apply the second part of the training rule using the results of Step 6.
Steps 1 through 3 are often called the forward pass, and steps 4 through 7 are often called
the backward pass. Hence, the name: back-propagation. For each data pair to be learned a
forward pass and backwards pass is performed. This is repeated over and over again until
the error is at a low enough level (or we give up).

Figure 5.1
Calculations and Transfer Function
120

The behaviour of a NN (Neural Network) depends on both the weights and the input-
output function (transfer function) that is specified for the units. This function typically
falls into one of three categories:
- linear
- threshold
- sigmoid
For linear units, the output activity is proportional to the total weighted output.
For threshold units, the output is set at one of two levels, depending on whether the total
input is greater than or less than some threshold value.
For sigmoid units, the output varies continuously but not linearly as the input changes.
Sigmoid units bear a greater resemblance to real neurons than do linear or threshold units,
but all three must be considered rough approximations.
It should be noted that the sigmoid curve is widely used as a transfer function because it
has the effect of "squashing" the inputs into the range [0,1]. Other functions with similar
features can be used, most commonly tanh which has an output range of [-1,1]. The
sigmoid function has the additional benefit of having an extremely simple derivative
function for backpropagating errors through a feed-forward neural network. This is how
the transfer functions look like:
121

To make a neural network performs some specific task, we must choose how the units are
connected to one another (see Figure 5.1), and we must set the weights on the
connections appropriately. The connections determine whether it is possible for one unit
to influence another. The weights specify the strength of the influence.
Typically the weights in a neural network are initially set to small random values; this
represents the network knowing nothing. As the training process proceeds, these weights
will converge to values allowing them to perform a useful computation. Thus it can be
said that the neural network commences knowing nothing and moves on to gain some
real knowledge.
To summarize, we can teach a three-layer network to perform a particular task by using
the following procedure:
122

I. We present the network with training examples, which consist of a pattern of
activities for the input units together with the desired pattern of activities for the
output units.
II. We determine how closely the actual output of the network matches the desired
output.
III. We change the weight of each connection so that the network produces a better
approximation of the desired output.
The advantages of using Artificial Neural Networks software are:
I. They are extremely powerful computational devices
II. Massive parallelism makes them very efficient.
III. They can learn and generalize from training data so there is no need for
enormous feats of programming.
IV. They are particularly fault tolerant this is equivalent to the graceful
degradation found in biological systems.
V. They are very noise tolerant so they can cope with situations where normal
symbolic systems would have difficulty.
VI. In principle, they can do anything a symbolic/logic system can do, and more.
Real life applications
The applications of artificial neural networks are found to fall within the following broad
categories:
Manufacturing and industry:
- Beer flavor prediction
- Wine grading prediction
- For highway maintenance programs
Government:
- Missile targeting
- Criminal behavior prediction
Banking and finance:
- Loan underwriting
- Credit scoring
- Stock market prediction
- Credit card fraud detection
- Real-estate appraisal
Science and medicine:
123

- Protein sequencing
- Tumor and tissue diagnosis
- Heart attack diagnosis
- New drug effectiveness
- Prediction of air and sea currents
In this book we will examine some detailed case studies with Excel spreadsheets
demonstrating how the MFNN has been successfully applied to problems as diverse as
- Credit Approval,
- Sales Forecasting,
- Predicting DJIA weekly prices,
- Predicting Real Estate value
- Classify Type of Flowers
This book contains 5 neural network models develop using Excel worksheets described
above. Instructions on how to build neural network model with Excel will be explained
step by step by looking at the 5 main sections shown below
a) Selecting and transforming data
b) the neural network architecture,
c) simple mathematic operations inside the neural network model
d) training the model and
e) using the trained model for forecasting

Let's start building:

1) The Credit Approval Model

Credit scoring is a technique to predict the creditworthiness of a candidate applying for a
loan, credit card, or mortgage. The ability to accurately predict the creditworthiness of an
applicant is a significant determinant of success in the financial lending industry.
Refusing credit to creditworthy applicants results in lost opportunity, while heavy
financial losses occur if credit is given indiscriminately to applicants who later default on
their obligations.

In this example, we will use neural network to forecast the risk level of granting a loan to
the applicant. It can be used to guide decisions for granting or denying new loan
applications.

Open the Workbook(Credit_Approval) in folder Chapter 5 and bring up worksheet (Raw
124

Data). Here we have 400 inputs patterns and desire outputs. There are 10 input factors
and 1 desire output (end result). We can see that, the data are still in alphabet form.
Neural network (NN) can only be fed with numeric data for training. So we need to
transform these raw data into numeric form.
This worksheet is self explanatory. For example, in column B (Input 2), we have the
marital status. NN cannot take or understand married or single. So we transform them
to 1 for married and 0 for single. We have to do this one by one manually. If you
select the worksheet(Transform Data), it contains exactly what has been transform from
worksheet(Raw Data).

Figure 5.2

Now, we can see that Column A to column L in the worksheet (Transform Data) are all
numerical. (see Figure 5.2 above) Apart from this transformation, we also need to
massage the numeric data a little bit. This is because NN will learn better if there is
uniformity in the data.
We need to transform all the 400 rows of data into range between the values 0 to 1. The
first 398 rows will be used as training data. The last 2 rows will be used for testing our
prediction later. Thus we need to scale all the data into the value between 0 to 1. How do
we do that?
1- Copy all data from Column A to L to Column N To Column Y
2- Then, select Scale Data on the nn_Solve menu (see Figure 5.3) (see Appendix A
on how to load nn_Solve.

125

Figure 5.3
Enter the reference that you want to scale in the Data Range. We scale Input 1 (Age)
first. Enter N12:N411 in the Data Range. Press the Tab key on your keyboard to exit.
When you press Tab, nn_Solve will automatically load the maximum (70) and the
minimum (15.83) in the Raw Data frame Min and Max textbox. (see Figure 5.4 below)

3- Then specify the maximum (1) and minimum (0) scale range. Click on the Scale
Now button. The raw data will be scaled.
nn_Solve will also automatically store the minimum (in cell N414) and the maximum
(cell N413) value of the raw data in the last row and first column of the raw data. (see
Figure 5.5 below)

Figure 5.4

126

Figure 5.5
The raw input data that need to be scale are Input 5 (Address Time), Input 6 (Job Time)
and Input 9 (Payment History). We do not need to scale Input 2,3,4,7,6,10 as these values
are within 0 to 1. We do not need to scale the desire output (Credit Risk) as the value is
already within 0 to 1.

Figure 5.6

Figure 5.6 above show the data after they have been scaled. We need the raw minimum
and maximum values later when we reverse the scale values back to raw value.

127

b) the neural network architecture
A neural network is a group of neurons connected together. Connecting neurons to form a
NN can be done in various ways. This worksheet; column N to column AK actually
contain the NN architecture shown below:

INPUT LAYER HIDDEN LAYER OUTPUT LAYER

Figure 5.7

128

There are 10 nodes or neuron on the input layer. 6 neurons on the hidden layer and 1
neuron on the output layer.
Those lines that connect the nodes are call weights. I only connect part of them. In reality
all the weights are connected layer by layer, like Figure 5.1 above
The number of neurons in the input layer depends on the number of possible inputs we
have, while the number of neurons in the output layer depends on the number of desired
outputs. Here we have 400 input patterns map to 400 desired or target outputs. We reserve
2 input patterns for testing later.
Like what you see from the Figure 5.7 above, this NN model consists of three layers:
1. Input layer with 10 neurons.
Column N = Input 1 (I1); Column O = Input 2 (I2); Column P = Input 3 (I3);
Column Q = Input 4 (I4) ; Column R = Input 5 (I5) ; Column S = Input 6 (I6);
Column T = Input 7 (I7); Column U = Input 8 (I8); Column V = Input 9 (I9);
Column W = Input 10 (I10)

2. Hidden layer with 6 neurons.
Column AD = Hidden Node 1 (H1); Column AE = Hidden Node 2 (H2);
Column AF = Hidden Node 3 (H3); Column AG = Hidden Node 4 (H4);
Column AH = Hidden Node 5 (H5); Column AI = Hidden Node 6 (H6)

3. Output layer with 1 neurons.
Column AK = Output Node 1
Now let's talk about the weights that connection all the neurons together
Note that:
The output of a neuron in a layer goes to all neurons in the following layer.
(in Fig 5.7, I only connect the weights between all the Input nodes to Hidden
Node 1)
We have 10 inputs node and 6 hidden nodes and 1 output nodes. Here the
number of weights are (10 x 6) + (6 x 1) = 66
Each neuron has its own input weights.
The output of the NN is reached by applying input values to the input layer,
passing the output of each neuron to the following layer as input.

I have put the weights vector in one column AA. So the weights are contain in cells:
From Input Layer to Hidden Layer
w(1,1) = $AA$12 -> connecting I1 to H1
129


and
so
on
w(2, 6) = $AA$63 -> connecting I2 to H6

From Hidden Layer to Output Layer

w(h1,1) = $AA$72 -> connecting H1 to O1
w(h2, 1) = $AA$73 -> connecting H2 to O1
130

After mapping the NN architecture to the worksheet and entering the input and desired
output data., it is time to see what is happening inside those nodes.

The number of hidden layers and how many neurons in each hidden layer cannot be well
defined in advance, and could change per network configuration and type of data. In
general the addition of a hidden layer could allow the network to learn more complex
patterns, but at the same time decreases its performance. You could start a network
configuration using a single hidden layer, and add more hidden layers if you notice that
the network is not learning as well as you like.
For this Credit Approval model, 1 hidden layer is sufficient. Select the cell AD12 (H1),
you can see Figure 5.8
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
131

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

132

actual output is 0.498473 in cell AK12. Since 0.495473 is not what we want, we minus
this with the desire output.
Error = AK12 AL12
After that we square the error to get a positive value
AM12 = AK12 AL12 ^ 2 ; we get 0.245494 (in AM12)
The closer the actual output of the network matches the desired output, the better. This is
only one pattern error. Since we have 398 rows of patterns, we fill down the formula
again until row 409 in this spreadsheet. Sum up all the error and take the average.
MSE = SUM(AM12:AM409)/398
Our objective is to minimize the MSE. We need to change the weight of each connection
so that the network produces a better approximation of the desired output.
In NN technical term, we call this step of changing the weights as TRAINING THE NEURAL
NETWORK.
In order to train a neural network to perform some task, we must adjust the weights of
each unit in such a way that the error between the desired output and the actual output is
reduced. This process requires that the neural network compute the error derivative of the
weights (EW). In other words, it must calculate how the error changes as each weight is
increased or decreased slightly. The back propagation algorithm is the most widely used
method for determining the EW.
The back-propagation algorithm is easiest to understand if all the units in the network are
linear. The algorithm computes each EW by first computing the EA, the rate at which the
error changes as the activity level of a unit is changed. For output units, the EA is simply
the difference between the actual and the desired output. To compute the EA for a hidden
unit in the layer just before the output layer, we first identify all the weights between that
hidden unit and the output units to which it is connected. We then multiply those weights
by the EAs of those output units and add the products. This sum equals the EA for the
chosen hidden unit. After calculating all the EAs in the hidden layer just before the output
layer, we can compute in like fashion the EAs for other layers, moving from layer to layer
in a direction opposite to the way activities propagate through the network. This is what
gives back propagation its name. Once the EA has been computed for a unit, it is straight
forward to compute the EW for each incoming connection of the unit. The EW is the
product of the EA and the activity through the incoming connection. Phew, what craps is
this back propagation!!!. Fortunately, you dont need to understand this, if you use MS
Excel Solver to build and train a neural network model.

d) Training NN as an Optimization Task Using Excel Solver
Training a neural network is, in most cases, an exercise in numerical optimization of a
usually nonlinear function. Methods of nonlinear optimization have been studied for
hundreds of years, and there is a huge literature on the subject in fields such as numerical
analysis, operations research, and statistical computing, e.g., Bertsekas 1995, Gill,
133

Murray, and Wright 1981. There is no single best method for nonlinear optimization. You
need to choose a method based on the characteristics of the problem to be solved.
MS Excel's Solver is a numerical optimization add-in (an additional file that extends the
capabilities of Excel). It can be fast, easy, and accurate.
For a medium size neural network model with moderate number of weights, various
quasi-Newton algorithms are efficient. For a large number of weights, various conjugate-
gradient algorithms are efficient. These two optimization method are available with Excel
Solver
To make a neural network that performs some specific task, we must choose how the
units are connected to one another, and we must set the weights on the connections
appropriately. The connections determine whether it is possible for one unit to influence
another. The weights specify the strength of the influence. Values between -1 to 1 will be
the best starting weights.
Let fill out the weight vector. The weights are contain in AA12:AA77. From the
nn_Solve menu, select Randomize Weights

Figure 5.10
Enter AA12:AA77 and click on the Randomize Weights button. AA12:AA77 will be
filled out with values between -1 to 1. (see Figure 5.11 below)
134

Figure 5.11
The learning algorithm improves the performance of the network by gradually changing
each weight in the proper direction. This is called an iterative procedure. Each iteration
makes the weights slightly more efficient at separating the target from the nontarget
examples. The iteration loop is usually carried out until no further improvement is being
made. In typical neural networks, this may be anywhere from ten to ten-thousand
iterations. Fortunately, we have Excel Solver. This tool has simplified neural network
training so much.

Accessing Excels Solver (see Appendix C to access Excels Solver for Excel 2007)

To use the Solver, click on the Tools heading on the menu bar and select the Solver . . .
item. (see Figure 5.12)

Figure 5.12
135

Figure 5.14

If Solver is not listed (see Figure 5.12) then you must manually include it in the
algorithms that Excel has available. To do this, select Tools from the menu bar and
choose the "Add-Ins . . ." item. In the Add-Ins dialog box, scroll down and click on the
Solver Add-In so that the box is checked as shown Figure 5.14 above:

After selecting the Solver Add-In and clicking on the OK button, Excel takes a moment
to call in the Solver file and adds it to the Tools menu.

If you cannot find the Solver Add-In, try using the Macs Find File or Find in Windows
to locate the file. Search for solver. Note the location of the file, return to the Add-Ins
dialog box (by executing Tools: Add-Ins), click on Select or Browse, and open the
Solver Add-In file.

What if you still cannot find it? Then it is likely your installation of Excel failed to
include the Solver Add-In. Run your Excel or Office Setup again from the original CD-
ROM and install the Solver Add-In. You should now be able to use the Solver by
clicking on the Tools heading on the menu bar and selecting the Solver item.

Although Solver is proprietary, you can download a trial version from Frontline Systems,
the makers of Solver, at www.frontsys.com.

After executing Tools: Solver . . . , you will be presented with the Solver Parameters
dialog box below:

136

Figure 5.15


cell. Here we enter cell AO1.

In our NN model, the objective function is to minimize the Mean Squared Error. See
Figure 5.16 below

Figure 5.16

137


Here, we select Min as we want to minimize the MSE


Here, the cells that need to be changed is the weights vector. In our model, the weights
are contain in Range AA12:AA77. So we enter, AA12:AA77. See Figure 5.17 below

Figure 5.17

Subject to the Constraints is used to impose constraints on the endogenous variables.
We will rely on this important part of Solver when we do Constrained Optimization
problems. Although, training a neural network is not a Constrained Optimization
problem, but it will be efficient if we limit the maximum and minimum value of the
weights to a range of -100 to 100. Therefore, we put in
138

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
139

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
140

Figure 5.21

Solve is obviously the button you click to get Excel's Solver to find a solution. This is the
last thing you do in the Solver Parameters dialog box. So, click Solve to start training.

Figure 5.22

spreadsheet. See Figure 5.22 above.

Figure 5.23

A message will appear after Solver has converged (see Figure 5.23)..In this case, Excel
This is good news!

141

Sometime, the Mean Square Error is not satisfactory and Solver unable to find the
solution at one go. If this is the case then you, Keep the Solver solution and run Solver
again. Follow the step discussed above. From experience, usually you will need to run
Solver a few times before Solver arrive at a satisfactory Mean Square Error. (Note: value
less than 0.01 will be satisfactory)

The two quickest fixes are to try different initial weights values and to add bigger or
smaller constraints to the weights.
Or you may change the network architecture by adding more hidden nodes.

to Keep the Solver Solution.

On the right-hand side of the Solver Results dialog box, Excel presents a series of reports.
The Answer, Sensitivity, and Limits reports are additional sheets inserted into the current
workbook. They contain diagnostic and other information and should be selected if
Solver is having trouble finding a solution.

It is important to understand that a saved Excel workbook will remember the information
included in the last Solver run.

Save Scenario... enables the user to save particular solutions for given configurations.

After all the training and the MSE is below 0.01, its now time for us to predict. Here, Ive
train the model and the MSE is 0.0086.

Goto the row 409 of the Credit Approval spreadsheet. Remember, we have save the last 2
rows for testing, row 410 and 411.

After that goto the hidden layer. Select AD409:AK409 (see Figure 5.24 below)

142

Figure 5.24

Fill down the formula to row 410 to 411. (see Figure 5.25)

Figure 5.25

On the Output cell i.e. AK410, the predicted value is 1 and AK411 is also 1
(see Figure 5.26 below). Thus, both results we got are High Risk. When you
compare to the actual result in L410 and L411, we are spot on. Thats it. You
have successfully use neural network to predict the risk involve when granting
a loan to an applicant.

143

Figure 5.26

144

2) The Sales Forecasting Model

Forecasting future retail sales is one of the most important activities that form the basis
for all strategic and planning decisions in effective operations of retail businesses as well
as retail supply chains.

Accurate forecasts of consumer retail sales can help improve retail supply chain
operation, especially for larger retailers who have a significant market share. For
profitable retail operations, accurate demand forecasting is crucial in organizing and
planning purchasing, production, transportation, and labor force, as well as after-sales
services. A poor forecast would result in either too much or too little inventory, directly
affecting the profitability of the supply chain and the competitive position of the
organization.

In this example, we will use neural network to forecast the weekly and daily sales of a
fashion store.

Open the Workbook(Sales_Forecasting) in folder Chapter 5 and bring up Worksheet (Raw
Data). There are 7 input factors and 2 desire output (end result). We also have 104 rows
of input patterns in this model.
We need to transform all the 104 rows of data into range between the values 0 to 1. The
first 102 rows will be used as training data. The last 2 rows will be used for testing our
prediction later. Here we can see that, the data are still in alphabet form. NN can only be
fed with numeric data for training. So we need to transform these raw data into numeric
form. (see Figure 5a.1) This worksheet is self explanatory. For example, in column B
(Input 2), we have the Season Influence. NN cannot take or understand Low, Medium,
High, Very High. So we transform them to Low = 0.25, Medium = 0.5, High = 0.75,
Very High = 0.9. We have to do this one by one manually.

145

Figure 5a.1
If you select the worksheet(Transform Data), it contains exactly what has been transform
from worksheet(Raw Data). (see Figure 5a.2)

Figure 5a.2

Now, we can see that Column A to column J in the worksheet (Transform Data) are all
numerical. Apart from this transformation, we also need to massage the numeric data a
little bit. This is because NN will learn better if there is uniformity in the data.
Thus we need to scale all the data into the value between 0 to 1. How do we do that?
Copy all data from Column A to J and paste them to Column L To U
146

Select Scale Data on the nn_Solve menu (Figure 5a.3)

Figure 5a.3

Figure 5a.4

Enter the reference for Input 1 (L7:L110) in the Data Range. Press the Tab key on your
keyboard to exit. When you press Tab, nn_Solve will automatically load the maximum
(104) and the minimum (1) in the Raw Data frame Min and Max textbox. (see Figure
5a.4)

Enter the value 1 for maximum and 0 for minimum for the Scale Into frame. Of course
you can change this.

Click on the Scale Now button. The raw data will be scaled
nn_Solve will also automatically store the minimum (in cell L113) and the maximum
(L112) value of the raw data in the last row and first column of the raw data. (see Figure
5a.5 below)

147

Figure 5a.5

We also need to scale Input 4 in column O, Input 5 in column P, Input 6 in column Q,
Input 7 in column R and the 2 Desire outputs in column T and U. Ive scale all the input
data for your convenience. See Figure 5a.6 below.

Figure 5a.6
We dont need to scale Input 2 and 3 as those values are already between 0 and 1.

148

NN can be done in various ways. This worksheet; column L to column AF actually


Figure 5a.7

neurons on the output layer.
all the weights are connected layer by layer, like Figure 5.1.

149

2 input patterns for testing later.
Like what you see from the Figure 5a.7 above, this NN model consists of three layers:
Input layer with 7 neurons.
Column L = Input 1 (I1); Column M = Input 2 (I2); Column N = Input 3 (I3);
Column O = Input 4 (I4) ; Column P = Input 5 (I5) ; Column Q = Input 6 (I6);
Column R = Input 7 (I7)

Hidden layer with 5 neurons.

Column Y = Hidden Node 1 (H1); Column Z = Hidden Node 2 (H2);
Column AA = Hidden Node 3 (H3); Column AB = Hidden Node 4 (H4)
Column AC = Hidden Node 5 (H5)

Output layer with 2 neurons.
Column AE = Output Node 1 (O1)
Column AF = Output Node 2 (O2)
Note that:
- The output of a neuron in a layer goes to all neurons in the following layer.
- In Figure 5a.7 we have 7 inputs node and 5 hidden nodes and 2 output nodes.
Here the number of weights are (7 x 5) + (5 x 2) = 45

- Each neuron has its own input weights.

- The output of the NN is reached by applying input values to the input layer,

I have put the weights vector in one column W. So the weights are contain in cells:
w(1,1) = $W$7 -> connecting I1 to H1

150

w(4,2) = $W$17-> connecting I4 to H2
and
so
on

w(2, 5) = $W$36 -> connecting I2 to H5


w(h1,1) = $W$42 -> connecting H1 to O1
w(h2, 1) = $ W$43 -> connecting H2 to O1
w(h4, 1) = $W$45 -> connecting H4 to O1

w(h1,2) = $W$47 -> connecting H1 to O2

151


For this Credit Approval model, 1 hidden layer is sufficient. Select the cell Y7 (H1), you
can see Figure 5a.8
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
152

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
153

MSE = (SUM(AJ7:AK108)/102)/2

Figure 5a.10
In NN technical term, we call this step of changing the weights as TRAINING THE
NEURAL NETWORK.

product of the EA and the activity through the incoming connection.
Phew, what craps is this back propagation!!!. Fortunately, you dont need to understand

154

this, if you use MS Excel Solver to build and train a neural network model.

Solver
units are connected to one another , and we must set the weights on the connections
Lets fill out the weight vector. The weights are contain in W7:W51. From the nn_Solve
menu, select Randomize Weights (see Figure 5a.11 below)

Figure 5a.11
Enter W7:W51 and click on the Randomize Weights button. W7:W51 will be filled out
with values between -1 to 1. (see Figure 5a.12 below)

155

Figure 5a.12
training so much .
Accessing Excels Solver
To invoke Solver see page 134. After executing Tools: Solver . . . , you will be presented
with the Solver Parameters dialog box below:

Figure 5a.16
156


cell. Here we enter cell AN1.

Figure 5a.17 below

Figure 5a.17

Figure 5a.18

Here, we select Min as we want to minimize the MSE

157

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

158

After that select the Options. This will allows you to adjust the way in which Solver
approaches the solution.. (see Figure 5a.21)

Figure 5a.21


For a neural network model, you can set :

i.Max Time: 1000 seconds
ii.Iterations: 10000
iii.Precision: 0.000001
iv.Tolerance: 5%
v.Convergence: 0.0001

the Mean Squared Error.


X
X
X
159

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
160

Figure 5a.24

A message will appear after Solver has converged (see Figure 5a.24). In this case, Excel
This is good news!

less than 0.01 will be very good)





After all the training and the MSE is below 0.01, its now time for us to predict. Goto the
row 109 of the Sales Forecasting spreadsheet. Remember, we have save 2 rows of data
for testing i.e. row 109 and 110.
161

Figure 5a.25

Then goto Y108. Select Y108:AF108. (see Figure 5a.26 below).

Figure 5a.26

After that, you fill down until row 110. (see Figure 5a.27)

Figure 5a.27

162

Figure 5a.28

So, for row 109 we have 0.903441 for predicted Output 1 (AE109) and 0.102692 for
predicted Output 2 (AF109). (see Figure 5a.28).
Row 110 we have 0.84799 for predicted Output 1 (AE110) and 0.080876 for predicted
Output 2 (AF110).
We need to scale these number back to raw data before they have any meaning to us.
Select Scale Data for the nn_Solve menu (see Figure 5a.29 below)

Figure 5a.29

163

Figure 5a.30

Enter AE109:AF110 in the Data Range. As we are reversing what we did just now when
we scale the raw data to the range 0 to 1. Now the raw data maximum become 1 and
minimum become 0.
As we have automatically save the maximum and minimum of the raw data initially when
we scale them, now we use them as the maximum and minimum values to be scaled into
(See above Figure 5a.30). Enter 369400 (in cell AG112) as the maximum and
16042.85714 (in cell AG113) as the minimum. Click on Scale Now.

Figure 5a.31
164

So our predicted first weekly sales is 335280 as in AE109 and first daily sales is
52329.66 as in AF109. And our predicted second weekly sales is 315686.2 as in
AE110 and second daily sales is 44620.95 as in AF110. (see Figure 5a.31 above)
The desire weekly sales is 364000 (see I109) and 353000 (see I110) respectively.
Whereas
desire daily sales is 52000 (see cell J109) and 50428.57 (see cell J110). The predicted
values that we have are within 10% error tolerance of these desire values. So it is
acceptable.
Of course when you do the training on your own, you will get slightly different result
because of the MSE error that you have derived. The results here are based on the
MSE of 0.00391.
There you go. You have successfully use neural network to predict the weekly and
daily sales.

165

3) Predicting the DJIA weekly price.

Neural networks are an emerging and challenging computational technology and they
offer a new avenue to explore the dynamics of a variety of financial applications. They
can simulate fundamental and technical analysis methods using fundamental and
technical indicators as inputs. Consumer price index, foreign reserve, GDP, export and
import volume, etc., could be used as inputs. For technical methods, the delayed time
series, moving averages, relative strength indices, etc., could be used as inputs of neural
networks to mine profitable knowledge.

In this example, I have built a neural network to forecast the weekly prices of the DJIA
by using the moving averages.


Open the Workbook(Dow_Weekly) in folder Chapter 5 and bring up worksheet (Raw
Data). Here, a real life example is use. This data is taken from the DJIA weekly prices,
from the period from 22 April 2002 to 15 Oct 2007. There are 5 input factors and 1 desire
output (end result), The input factors consist of 5-days Moving Average, 10-days Moving
Average, 20-days Moving Average, 60-days Moving Average and 120-days Moving
Average. The desire output is the next week DJIA price. There are 287 rows of input
patterns in this model. (see Figure 5b.1a)

Figure 5b.1a

There are 287 rows of data. The first 286 rows will be used as training data. The last 1
row will be used for testing our prediction later.
We need to massage the numeric data a little bit. This is because NN will learn better if
there is uniformity in the data.
166

Goto worksheet(Transform Data).Copy all data from Column B to G and paste them to
Column J To O.
Select Scale Data on the nn_Solve menu. (see Figure 5b.1)

Figure 5b.1

Thus we need to convert the data in the range J3:O289 (see Figure 5b.2 ). That is 6
columns times 287 rows in the worksheet (Transform Data). Enter J3:O289 into the Data
Range edit box. Press the Tab key on your keyboard to exit. When you press Tab,
nn_Solve will automatically load the maximum (14,093.08) and the minimum (7528.4)
in the Raw Data frame Min and Max textbox.

Enter the value 1 for maximum and 0.1 for minimum for the Scale Into. Of course you
can change this. It is advisable not to enter the value 0 as the minimum as it represent
nothing. (see Figure 5b.2 below)

Figure 5b.2

167

Click on the Scale Now button, nn_Solve will also automatically store the minimum (in
cell J292) and the maximum (cell J291) value of the raw data in the last row and first
column of the raw data. (see Figure 5b.3 below). We may need these numbers later. Ive
convert all the values for you already in the worksheet (Transform Data).

Figure 5b.3

After converting the values, its now time to build the neural network infrastructure.

168

NN can be done in various ways. This worksheet; column J to column X actually contain
the NN architecture shown below:

Figure 5b.4
neuron on the output layer.
all the weights are connected layer by layer, like Figure 5.1
1 input pattern for testing later.
Like what you see from the Figure 5b.4 above, this NN model consists of three layers:
Column J = Input 1 (I1); Column K = Input 2 (I2); Column L = Input 3 (I3);
Column M = Input 4 (I4) ; Column N = Input 5 (I5)

Column T = Hidden Node 1 (H1)
Column U = Hidden Node 2 (H2)
169

Column V = Hidden Node 3 (H3)

Output layer with 1 neuron.
Column X = Output Node 1 (O1)

Note that:
- The output of a neuron in a layer goes to all neurons in the following layer.
- We have 5 inputs node and 3 hidden nodes and 1 output node. Here the number of
weights are (5 x 3) + (3 x 1) = 18
- Each neuron has its own input weights.
- The output of the NN is reached by applying input values to the input layer,

I have put the weights vector in one column Q. So the weights are contain in cells:
w(1,1) = $Q$3 -> connecting I1 to H1

w(4,2) = $Q$11-> connecting I4 to H2
and
so
on

w(2, 5) = $Q$14 -> connecting I2 to H5
170



w(h1,1) = $Q$18 -> connecting H1 to O1
w(h2, 1) = $ Q$19 -> connecting H2 to O1
w(h3, 1) = $Q$20 -> connecting H3 to O1

For this model, 1 hidden layer is sufficient. Select the cell T3 (H1), you can see
X
X
X
X
X
X
X
X
X
X
X
X
X
171

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
172

Figure 5b.5 above indicate that the desire Output 1 is 0.4397587 i.e. Y3. The actual
output is 0.529467 i.e. X3. Since 0.529467 is not what we want, we minus this with the
desire output. After that we square the error to get a positive value
Error = (Y3 X3) ^ 2 ; we get 0.008048 (in AA3)
The closer the actual output of the network matches the desired output, the better. This is
only one pattern error. Since we have using 286 rows of patterns, we fill down the
formula again from row 3 until row 288 in this spreadsheet. Remember, we save the last 1
row for testing our model later. Sum up all the error and take the average.
MSE = (SUM(AA3:AA288)/286)
so that the network produces a better approximation of the desired output. In NN
technical term, we call this step of changing the weights as TRAINING THE NEURAL
NETWORK.
method for determining the EW. The back-propagation algorithm is easiest to understand
if all the units in the network are linear. The algorithm computes each EW by first
computing the EA, the rate at which the error changes as the activity level of a unit is
changed. For output units, the EA is simply the difference between the actual and the
desired output. To compute the EA for a hidden unit in the layer just before the output
layer, we first identify all the weights between that hidden unit and the output units to
which it is connected. We then multiply those weights by the EAs of those output units
and add the products. This sum equals the EA for the chosen hidden unit. After
calculating all the EAs in the hidden layer just before the output layer, we can compute in
like fashion the EAs for other layers, moving from layer to layer in a direction opposite to
the way activities propagate through the network. This is what gives back propagation its
name. Once the EA has been computed for a unit, it is straight forward to compute the
EW for each incoming connection of the unit. The EW is the product of the EA and the
activity through the incoming connection.

173

capabilities of Excel). It can be fast, easy, and accurate. For a medium size neural
network model with moderate number of weights, various quasi-Newton algorithms are
efficient. For a large number of weights, various conjugate-gradient algorithms are
efficient. These two optimization method are available with Excel Solver. To make a
neural network that performs some specific task, we must choose how the units are
connected to one another, and we must set the weights on the connections appropriately.
The connections determine whether it is possible for one unit to influence another. The
weights specify the strength of the influence. Values between -1 to 1 will be the best
starting weights. Let fill out the weight vector. The weights are contain in Q3:Q20. From
the nn_Solve menu, select Randomize Weights (see Figure 5b.6)

Figure 5b.6
Enter Q3:Q20 and click on the Randomize Weights button. Q3:Q20 will be filled out with
values between -1 to 1. (see Figure 5b.7 below)

Figure 5b.7
174

training so much.
To invoke Solver goto page 134. After executing Tools: Solver . . . , you will be
presented with the Solver Parameters dialog box below:

Figure 5b.11

cell. Here we enter cell AD1.

In our NN model, the objective function is to minimize the Mean Squared Error (AD1).
See Figure 5b.12 and 5b.13 below

X
X
X
X
X
X
X
175

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
176

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
177

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
178

Figure 5b.19

spreadsheet. See Figure 5b.19 above.

Figure 5b.20

A message appears when Solver converged (see Figure 5b.20). In this case, Excel reports
that Solver has converged to the current solution. All constraints are satisfied. This is
good news!



179


row 288 of the Dow_Weekly spreadsheet. Remember, we have save 1 row of data for
testing i.e. row 289

Figure 5b.21

Then goto T288. Select T288:X288. See Figure 5b.22 below.

Figure 5b.22
After that, you fill down until row 289 (see Figure 5b.23 below)

.
Figure 5b.23
180

So, for row 109 we have 0.946003 for predicted Output 1 (X289).
We need to scale this number back to raw data before they have any meaning to us.
Select Scale Data from the nn_Solve menu.

Figure 5b.24
Enter X289 in the Data Range. As we are reversing what we did just now when we scale
the raw data to the range 0 to 1. Now the raw data maximum become 1 and minimum
become 0.1 (see Figure 5b.24)
As we have save the maximum (14093.080 and minimum (7528.4) of the raw data, now
we use them as the maximum and minimum values to be scaled into. Click on Scale Now.

Figure 5b.25

181

So our predicted DJIA weekly price is 13860.05 as in X289 (see Figure 5b.25). The
actual price is 13879.39 (cell G289). Of course when you do the training on your own,
you will get slightly different result because of the MSE error that you have derived.
The results here are based on the MSE of 0.001265
There you go. You have successfully use neural network to predict the next week
price of the DJIA.. You can also use other factors as inputs data. For example, like
Volume, Bollinger bands, RSI, ADX and etc. The sky is the limit and use your
creativity.

4) Predicting Real Estate Value

Our objective is to use neural network to forecast the value of a residential property in a
suburban area.

Open the workbook(Real_Estate) in Chapter 5 and bring up worksheet (Raw Data). Here
we have 499 inputs patterns and desire outputs. There are 13 input factors and 1 desire
output (end result). Goto the worksheet (Description) to see the explanation of each of the
input factors.
We need to massage these numeric data a little bit. This is because NN will learn better
if there is uniformity in the data. Goto the worksheet (Transform Data)
Copy all data from Column A to O and paste them to Column Q To AE. The first column
we need to scale is Column Q (Input 1)
Select Scale Data on the nn_Solve menu

Figure 5c.1
Enter the reference for Input 1 (Q5:Q503) in the Data Range. Press the Tab key on your
(88.9762) and the minimum (0.00632) in the Raw Data frame Min and Max textbox. (see
Figure 5c.2)

182

Figure 5c.2
Enter the value 1 for maximum and 0.1 for minimum for the Scale Into. Of course you
can change this. It is advisable not to use the value 0 as the minimum as it represent
nothing. Click on the Scale Now button. The raw data will be scaled.
nn_Solve will also automatically store the minimum (in cell Q506) and the maximum
(Q505) value of the raw data in the last row and first column of the raw data. (see Figure
5c.3 below)

Figure 5c.3
We also need to scale Input 2 in column R, Input 3 in column S, Input 6 in column V,
183

Input 7 in column W, Input 8 in column X, Input 9 in column Y, Input 10 in column Z,
Input 11 in column AA, Input 12 in column AB, Input 13 in column AC and Desire
Output in Column AE. Ive scale all the input data for your convenience. See Figure 5c.4
below
We dont need to scale Input 4 in column T and Input 5 in column U as those values are
already between 0 and 1.

Figure 5c.4

NN can be done in various ways. This worksheet; column Q to column AR actually
184


Figure 5c.5

There are 13 (column Q:AE) nodes or neuron on the input layer. 7 neurons on the hidden
layer (column AJ:AP) and 1 neuron on the output layer (column AR).


185

Like what you see from the Figure 5c.5 above, this NN model consists of three layers:
Column Q = Input 1 (I1); Column R = Input 2 (I2); Column S = Input 3 (I3);
Column T = Input 4 (I4) ; Column U = Input 5 (I5) ; Column V = Input 6 (I6);
Column W = Input 7 (I7); Column X = Input 8 (I8); Column Y = Input 9 (I9);
Column Z = Input 10 (I10); Column AA = Input 11 (I11); Column AB = Input 12 (I12);
Column AC = Input 13 (I13)


Column AJ = Hidden Node 1 (H1); Column AK = Hidden Node 2 (H2);
Column AL = Hidden Node 3 (H3); Column AM = Hidden Node 4 (H4)
Column AN = Hidden Node 5 (H5); Column AO = Hidden Node 6 (H6)
Column AP = Hidden Node 7 (H7)

Column AR = Output Node 1 (O1)
Note that:
o The output of a neuron in a layer goes to all neurons in the following layer.
See Figure 5c.5

o We have 13 inputs node and 7 hidden nodes and 1 output node. Here the
number of weights are (13 x 7) + (7 x 1) = 98

o Each neuron has its own input weights.

o The output of the NN is reached by applying input values to the input
layer, passing the output of each neuron to the following layer as input.

I have put the weights vector in one column. So the weights are contain in cells:
w(1,1) = $AG$5 -> connecting I1 to H1
186


w(4,2) = $AG$21-> connecting I4 to H2

and
so
on
187



w(h1,1) = $AG$96 -> connecting H1 to O1
w(h2, 1) = $AG$97 -> connecting H2 to O1


For this Real Estate model, 1 hidden layer is sufficient. Select the cell AJ5 (H1), you can
see
X
X
X
X
X
X
188

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
189

X

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

190

In NN technical term, we call this step of changing the weights as TRAINING THE
NEURAL NETWORK.
product of the EA and the activity through the incoming connection. Phew, what craps is
this back propagation!!!. Fortunately, you dont need to understand this, if you use MS
Excel Solver to build and train a neural network model.

Solver
191

Let fill out the weight vector. The weights are contain in AG5:AG102. From the
nn_Solve menu, select Randomize Weights (see Figure 5c.9)

Figure 5c.9
Enter AG5:AG102 and click on the Randomize Weights button. AG5:AG102 will be
filled out with values between -1 to 1. (see Figure 5c.10 below)
training so much.

Figure 5c.10
192



Figure 5c.14


cell. Here we enter cell AV1.

Figure 5c.15 below

193

Figure 5c.15

Here, we select Min as we want to minimize MSE


are contain in Range AG5:AG102. So we enter, AG5:AG102. (See Figure 5c.15 above)


AG5:AG102 > -100 and
AG5:AG102< 100

194

Click on the Add button to add these constraints. (see Figure 5c.16)

Figure 5c.16

Figure 5c.17
After that select the Options. This will allows you to adjust the way in which Solver
approaches the solution.. (see Figure 5c.18 below)

195

Figure 5c.18


For a neural network model, you can set :

i.Max Time: 1000 seconds
ii.Iterations: 10000
iii.Precision: 0.000001
iv.Tolerance: 5%
v.Convergence: 0.0001

the Mean Squared Error. (see Figure 5c.18 above)



X
X
X
X
196

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
197

A message appears when Solver converged (see Figure 5c.21). In this case, Excel reports
good news!






Save Scenario... enables the user to save particular solutions for given configurations.

row 502 of the RealEstate spreadsheet. Remember, we have save 1 row of data for testing
i.e. row 503

198

Figure 5c.22

Select AJ502:AR502. (See Figure 5c.22 above)

After that, you fill down until row 503 (see Figure 5c.23 below)

.
Figure 5c.23

So, for row 503 we have 0.26748 for predicted Output 1 (AR503). (see Figure 5c.23)
We need to scale this number back to raw data before they have any meaning to us.
Select Scale Data for the nn_Solve menu.

Figure 5c.24

199

Figure 5c.25

Enter AR503 in the Data Range. As we are reversing what we did just now when we
scale the raw data to the range 0 to 1. Now the raw data maximum become 1 and
minimum become 0.1
As we have automatically saved the maximum (50) and minimum (5) of the raw data
initially, now we use them as the maximum and minimum values to be scaled into. (See
Figure 5c.25 above). Click on Scale Now.

Figure 5c.26

So our predicted real estate price is 17.0366 as in AR503 (see Figure 5c.26 above). The
actual price is 17.2 (cell O503). Of course when you do the training on your own, you
will get slightly different result because of the MSE error that you have derived. The
results here are based on the MSE of 0.00178
There you go. You have successfully use neural network to predict real estate price.

200

5) Classify Type Of Irises

Open the file Irises.xls. Using the petal and sepal sizes, we can use neural network to
classify which class an Iris flower belongs to.

This is one of the standard benchmark that can be used to show how neural networks (and
other techniques) can be used for classification. The neural network is train with 146
examples of three species of Iris. We reserve 1 example for testing our model later. So we
have 147 records altogether. Two of the species are not linearly separable, so there is no
simple rule for classification of flowers. After proper training the network is capable of
classifying the flowers with a 100% accuracy.

Open the Workbook(Irises) in folder Chapter 5 and bring up worksheet (Raw Data). Here
we, 147 inputs patterns and desire outputs. There are 4 input factors and 1 desire output
(end result). Here we can see that, the data are still in alphabet form. NN can only be fed
with numeric data for training. So we need to transform these raw data into numeric form.
This worksheet is self explanatory. For example, in column F (Desire), we have the
Flower Type. NN cannot take or understand setosa, versicol, virginic. So we transform
them to 1 for setosa, 0.5 for versicol and 0 for virginic.
We have to do this one by one manually.
If you select the worksheet(Transform Data), it contains exactly what has been transform
from worksheet(Raw Data). (see Figure 5d.1)

Figure 5d.1

201

Now, we can see that Column A to column F in the worksheet (Transform Data) are all
numerical. Apart from this transformation, we also need to massage the numeric data a
little bit. This is because NN will learn better if there is uniformity in the data.
Copy all data from Column A to F and paste them to Column H to M. The first column
we need to scale is Column H (Input 1)
Select Scale Data on the nn_Solve menu (see Figure 5d.2)

Figure 5d.2

Figure 5d.3

202

Enter the reference for Input 1 (H7:H153) in the Data Range. Press the Tab key on your
(7.9) and the minimum (4.3) in the Raw Data frame Min and Max textbox. (see Figure
5d.2). Enter the value 1 for maximum and 0.1 for minimum for the Scale Into. Of course
you can change this. It is advisable not to use the value 0 as the minimum as it represent
nothing. Click on the Scale Now button. The raw data will be scaled.
nn_Solve will also automatically store the minimum (in cell H156) and the maximum
(H155) value of the raw data in the last row and first column of the raw data. (see Figure
5d.4 below). We may need these numbers later.

Figure 5d.4
We also need to scale Input 2 in column I, Input 3 in column J and Input 4 in column K.
Ive scale all the input data for your convenience. See Figure 5d.5 below.
We dont need to scale Desire column M as those values are already between 0 and 1.

Figure 5d.5
203

NN can be done in various ways. This worksheet; column H to column X actually contain
the NN architecture shown below: It is a 4 layer neural network which is very effective in
classification problems.

INPUT LAYER HIDDEN LAYER 1 HIDDEN LAYER 2 OUTPUT LAYER

Figure 5d.6

There are 4 (column H:K) nodes or neuron on the input layer. 2 neurons on the hidden
layer 1 (column R:S), 2 neurons on the hidden layer 2 (column U:V) and 1 neuron on the
output layer (column X).


Like what you see from the Figure 5d.6 above, this NN model consists of three layers:
204


Column H = Input 1 (I1); Column I = Input 2 (I2); Column J = Input 3 (I3);
Column K = Input 4 (I4)

Hidden layer 1 with 2 neurons.

Column R = Hidden Node 1 (H1)
Column S = Hidden Node 2 (H2)

Hidden layer 2 with 2 neurons.

Column U = Hidden Node 1 (H3)
Column V = Hidden Node 2 (H4)

Column X = Output Node 1 (O1)
Note that:
o The output of a neuron in a layer goes to all neurons in the following layer.
See Figure 5d.6

o We have 4 inputs node and 2 nodes for hidden layer 1, 2 nodes for hidden
layer 2 and 1 output node. Here the number of weights are (4 x 2) + (2 x 2)
+ (2 x 1) = 14

o Each neuron has its own input weights.

o The output of the NN is reached by applying input values to the input
layer, passing the output of each neuron to the following layer as input.

I have put the weights vector in one column O. So the weights are contain in cells:
From Input Layer to Hidden Layer 1
w(1,1) = $O$7 -> connecting I1 to H1

205

w(4,2) = $O$14-> connecting I4 to H2
and
so
on

From Hidden Layer 1 to Hidden Layer 2
w(h1,h3) = $O$15 -> connecting H1 to H3
w(h2,h4) = $O$18-> connecting H2 to H4

From Hidden Layer 2 to Output Layer

w(h3,1) = $O$19 -> connecting H3 to O1
w(h4,1) = $O$20 -> connecting H4 to O1


For this Irises classification model, 2 hidden layers are needed. Select the cell R7 (H1),
you can see
X
X
X
X
X
X
206

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
207

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

208

product of the EA and the activity through the incoming connection.

need to choose a method based on the characteristics of the problem to be solved. MS
Excel's Solver is a numerical optimization add-in (an additional file that extends the
Solver
Let fill out the weight vector. The weights are contain in O7:O20. From the nn_Solve
menu, select Randomize Weights (see Figure 5d.8)
209

Figure 5d.8
Enter O7:O20 and click on the Randomize Weights button. O7:O20 will be filled out with
values between -1 to 1. (see Figure 5d.9 below)

Figure 5d.9
makes the weights slightly more efficient at separating the target from the non target
training so much.
210



Figure 5d.13


cell. Here we enter cell AB1.

Figure 5d.14 below

211

Figure 5d.14

Here, we select Min as we want to minimize MSE


are contain in Range O7:O20. So we enter, O7:O20. (See Figure 5d.14 above)


O7:O20 > -100 and
O7:O20 < 100

212

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
213

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
214

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
215

A message appears when Solver converged (see Figure 5d.20). In this case, Excel reports
good news!





row 152 of the Irises spreadsheet. Remember, we have save 1 row of data for testing i.e.
row 153

Figure 5d.21

216

Select R152:X152. (See Figure 5d.21 above).After that, you fill down until row 153 (see
Figure 5d.22 below)

.Figure 5d.22

So, for row 153 we have 0.08839486 for predicted Output 1 (X153). (see Figure 5d.23
below). As the value 0.08839486 is very near to 0, we can take this result as 0.

Figure 5d.23
So our predicted type of Irises price is Virginic which is represented by the value 0 as in
X153 (see Figure 5d.23 above). The desire output is 0 (cell M153). We have successfully
predicted the exact outcome. Of course when you do the training on your own, you will
get slightly different result because of the MSE error that you have derived. The results
here are based on the MSE of 0.00878
The general rules for classification are, when you have a predicted value of less than 0.4,
this can be round up to the value of 0. And if you have a predicted value of more than 0.6,
then you can round up to the value of 1. My suggestion is, try to rebuild and retrain the
NN model if you get borderline cases like this.
There you go. You have successfully use neural network to predict the type of Irises.

217

Conclusion

Neural networks tolerance to noise makes them an excellent choice for solving real-
world pattern recognition problems in which perfect data is not always available. As with
any solution, there are costs. Training a network can be an arduous process. Depending
on the domain, obtaining sufficient and suitable training data, sometimes called "truth",
can be challenging. For example, speech transcription and natural language systems
require large amounts of data to train. In addition, after the system is implemented it may
not converge. That is, regardless of the amount of training, the weights may not always
"settle" at particular values. In these cases, developing neural networks becomes more of
an art than a science.

Each problem is unique, and so too are the solutions. Sometimes adding additional nodes
or layers will stabilize a system. There are no hard rules, but there is one thing for certain;
whether a neural network is computed by a computer, implemented in hardware, or
propagated by hand, neural networks do not cogitate. They are simply powerful
computational tools and their sophisticated mathematical computation positions them to
be used to solve a broad range of problems.

Neural networks are increasingly being used in real-world business applications and, in
some cases, such as fraud detection, they have already become the method of choice.
Their use for risk assessment is also growing and they have been employed to visualize
complex databases for marketing segmentation. This boom in applications covers a wide
range of business interests from finance management, through forecasting, to
production. With this book, you have learned the neural network method to enable direct
quantitative studies to be carried out without the need for rocket-science expertise.

218

Chapter 6

Markov Chain

Introduction

MARKOV CHAIN, a powerful term but simple to understand. Markov chains are used to
analyze trends and predict the future (weather, stock market, genetics, product success,
etc.)

A Markov Chain is a special kind of stochastic process where the outcome of an experiment
depends only on the outcome of a previous experiment. In other words, the next state of the
system depends only on the current state and not on the previous states. Stochastic processes
are of interest for describing the behaviour of a system evolving over a period of time.

Suppose there is a physical or mathematical system that has n possible states and at any
one time, the system is in one and only one of its n states. The state of the system at time
t+1 depends only on the state of the system at time t. Such a system is called Markov
Chain or Markov process.

Figure 6.1

If a Markov process meets the following conditions:
1. The system can be described by a set of finite states and that the system can be in one
and only one state at a given time.
2 The transition probability Pr
ij
, the probability of transition from state i to state j, is
given from every possible combination of i and j (including i = j) and the transition
probabilities are assumed to be stationary (unchanging) over the time period of interest
and independent of how state i was reached. and
3. Either the initial state of the system or the probability distribution of the initial state is
known.
Then, is called a finite-state first order Markov chain.

To understand Markov Chains and do analytical calculations with Markov Chains, you
must
have a basic understanding of probability and matrices. For the example in the activity on
sunny and rainy days, you will write the probability of one day being rainy and the
probability of one day being sunny. You should be able to take that probability and assess
219

how many sunny or rainy days there are in a year. Ability in matrix theory isn't necessary
to complete this lesson, but is necessary if you wish to scale this lesson up, move beyond
simulations, and do analytical calculations. Below is a refresher on how to do matrix
multiplication.

An example of a Markov chain is a string with 2 nodes on it. These nodes are called node
0
and node 1. If we are at node 0 we can stay at node 0 with probability .1 and move to
node 1
with probability .9. If we are at node 1 we can stay at node 1 with probability .7 or move
to node 0 with probability .3. Here is how we would set up our probability matrix.

We then select a set of 2 numbers between 0 and 1 that add to one and create a matrix
with
them as follows. Let us select .5 and .5. These numbers represent a guess of the
probability of being in node 0 and node 2.

We then multiply those two matrices together to get a new 2 x 1 matrix.

We multiply that onto the probability matrix again to get a new 2 x 1 matrix and we repeat until
we continue to get the same matrix. For our system we ultimately get this.

220

This represents that we spend 25% of our time on node 0 on the string, and 75% of our
time on node 1.

Lets start to further understand Markov Chain with the following example.

Example 1: The Courier Company
Suppose that a courier company has to deliver parcels from its customers in three
locations in New York: Manhattan location (labelled A), Brooklyn end location (labelled
B) and a Newark end location (labelled C). The company has a group of delivery drivers
to serve all three locations. The company's scheduler has determined the following:
1. Of the calls to the Manhattan location, 30% are delivered in Manhattan area, 30%
are delivered in the Brooklyn end, and 40% are delivered in the Newark end
2. Of the calls to the Brooklyn end location, 40% are delivered in Manhattan area,
40% are delivered in the Brooklyn end, and 20% are delivered in the Newark end
3. Of the calls to the Newark end location, 50% are delivered in Manhattan area,
30% are delivered in the Brooklyn end, and 20% are delivered in the Newark end.
After a delivery is made, a driver will go to the nearest location to make the next delivery.
Therefore, the location of a specific driver is determined only by his or her previous
location.
The delivery can be modelled with the following matrix:

Figure 6.2

0.3 = 30% ; 0.2 = 20% ; 0.4 = 40%
T is called the transition matrix of the above system. In our example, a state is the
location of a particular driver in the system at a particular time. The entry s
ji
in the above
matrix represents the probability of transition from the state corresponding to i to the state
corresponding to j. (e.g. the state corresponding to 2 is B)
221

To put this in a simple way, let's assume that it takes each delivery person the same
amount of time (say 20 minutes) to make a delivery, and then to get to their next location.
According to the statistician's data, after 20 minutes, of the drivers that began in A, 30%
will again be in A, 30% will be in B, and 40% will be in C. (see Figure 6.3 below)

Figure 6.3
Since all drivers are in one of those three locations after their delivery, each column sums
to 1; that is 0.3 + 0.3 + 0.4 = 1. As this involves the theory of probabilities, each entry
must be between 0 and 1, inclusive. Based on this important fact we can model this
situation as a Markov chain where the next location for delivery depends only on the
current location, not previous history. It is also true that our matrix of probabilities does
not change during the time we are observing.
Now, lets elaborate this further with a simple question. If you start at location C,
calculate the probability (say, P) that you will be in area B after 2 deliveries?
Accordingly we can get to B in two steps.
We can go from C to C, then from C to B, we can go from C to B, then from B to B,
or
we can go from C to A, then from A to B. To calculate P, let P(XY) represent the
probability of going from X to Y in one delivery (where X,Y can be A,B or C).
Before we go further, lets refresh on how probabilities work. If two (or more)
independent events must both (all) happen, to obtain the probability of them both (all)
happening, we multiply their probabilities together. To get the probability of either (any)
happening, we add the probabilities of those events together.
222

Figure 6.3a
This gives us
P = P(CA)P(AB) + P(CB)P(BB) + P(CC)P(CB) for the probability that a delivery person
goes from C to B in 2 deliveries.
CA = 0.5, AB =0.3, CB = 0.3, BB=0.4, CC = 0.2, CB = 0.3 (see Figure 6.3a above)
Substituting into our formula using the statistician's data above gives
P = (.5)(.3) + (.3)(.4) + (.2)(.3) = .33
Thus if we start at location C, we have a 33% chance of being in location B after 2
deliveries.
Let's look at another pair. If we start at location B, what is the probability of being at
location B after 2 deliveries? Using the same steps shown above, the probability of going
from location B to location B in two deliveries is
P(BA)P(AB) + P(BB)P(BB) + P(BC)P(CB) = (.4)(.3)+(.4)(.4) + (.2)(.3) = .34.
Its easy to calculate after 2 deliveries, but what if you need to know where you will be
after 5, or 15 deliveries? Its going to take quite a long time. Fortunately there is an
easier method. When you look carefully at these numbers you can see they are the result
of matrix multiplication.. Going from C to B in 2 deliveries is the same as taking the
inner product of row 2 and column 3 (row 2 multiply by row 3) Going from B to B in 2
deliveries is the same as taking the inner product of row 2 and column 2 (row 2 multiply
by column 2). If you multiply T by T, the (2, 3) and (2,2) entries are respectively, the
same answers that you got for these two questions above. The rest of T
2
answers the same
type of question for any other pair of locations X and Y.

223

Figure 6.4
As you can see the elements on each column still add to 1 and each element is between 0
and 1, inclusive. This is a requirement since we are modelling our problem with a
Markov chain. This matrix indicates the probabilities of going from location i to location
j in exactly 2 deliveries.
With this matrix, it is much easier to find where we will be after 3 deliveries. We will let
p(AB) represent the probability of going from A to B in 2 deliveries.
Calculate the probability of going from C to B in 3 deliveries: It is
p(CA)P(AB) + p(CB)P(BB) + p(CC)P(CB) = (.37)(.3) + (.33)(.4) + (.3)(.3) = .333
You will see that this probability is the inner product of row 2 of T
2
and column 3 of T.
Therefore, if we multiply T
2
by T, we will get the probability matrix for 3 deliveries.

With the examples shown above you should know how we find the matrix of
probabilities for 4, 5 or more deliveries. Just keep on multiplying the T matrix. You will
find out that the elements on each column still add to 1. It is vital that you keep as many
decimal places as possible to retain accuracy. By any means do not round your answers.

As we calculate more and more deliveries you will see a pattern in these matrices. The
numbers in each row seems to be converging to a particular number. It can be observed
that the difference between the elements of the transition probabilities after a number of
transitions tends to be zero. Think about what this tells us about our long-term
probabilities. This tells us that after a large number of deliveries, it no longer matters
which location we were in when we started. At the end of the week, we have
(approximately) a 38.9% Chance of being at location A, a 33.3% chance of being at
224

location B, and a 27.8% chance of being in location C. This convergence will happen
with most of the transition matrices that we consider.
In other words all elements of all rows tend towards a common limit as the number of
transition increases. Thus the process is said to have reached a steady state

I know that it is quite cumbersome to calculate the T matrix. Luckily there is a built in
formula in MS Excel that you can use for matrix multiplication. The formula is
=MMULT(). Open the file Markov.xlsx in the folder Chapter 6. Select worksheet
Transport

X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
x

Example 2: The Cola Wars
225

Assume there are three soft drink companies, King, Peps, and Royal, which have
consistent customer preference from year to year represented by the following transition
matrix T. (see Figure 6.7)

Figure 6.7

Given that a persons last cola purchase was King cola, there is a 75% chance that his
next cola purchase will also be King cola, 10% will be Peps and 15% will be Royal.
Given that a persons last cola purchase was Peps cola, there is a 55% chance that his
next cola purchase will also be Peps cola, 33%% will be King and 12% will be Royal.
Given that a persons last cola purchase was Royal cola, there is a 88% chance that his
next cola purchase will also be Royal cola, 10%% will be King and 2% will be Peps.
Lets look at a few scenarios for further understanding of Markov Chain.

Scenario 1:
Given that a person is currently a King cola purchaser, what is the probability that he will
purchase Peps cola two purchases from now?

Select the worksheet Cola. The formulas are entered like the one explained above for
worksheet Transport.
Two purchase from now will be T x T x T (T^3)

The value is 0.13933 i.e. there is a 13.933% that the next two purchases from now will be
a Peps cola (see Figure 6.8)

X
X
X
X
X
X
X
X
X
X
X
X
X
X
226

Figure 6.9
Scenario 3:
Assume each person makes one cola purchase per week. Suppose 50% of all people now
drink King, 30% drink Peps and 20% drink Royal. What fraction of people will be
drinking King cola three weeks from now?
So here we have the market shares of the three Colas which is also the initial states of 0.5,
0.3 and 0.2 respectively

Thus we will use the matrix T^4 for our calculation. (3 weeks means next three purchases
from now)

Pr[X3=King] = 0.5 x 0.47487 + 0.3 x 0.45574 + 0.2 x 0.25485 = 0.42513

Thus the fraction of people will be drinking King cola three weeks from now is 42.513%
I hope the above examples gave you a good idea about the process of Markov chains.
There is a rich mathematical theory of Markov chain (unless you have a mathematical
bent, you will probably want to skip heavy stuff below.) The important objective of this
book is to show readers on how to build a prediction model practically with MS Excel.
However if you want to know the theoretical aspect of Markov Chain, please read on.
Definitions
From a mathematical point of view, a Markov chain describes a process that can be
considered to be in exactly one of a number of "states" at any given time. A Markov
chain with n states, the state vector is a column vector whose i
th
component represents
the probability that the system is in the i
th
state at that time.
The sum of the entries of a state vector must be 1. For example, vectors X
0
and X
1
in the
above example are state vectors. If p
ij
is the probability of movement (transition) from
one state j to state i, then the matrix T=[ p
ij
] is called the transition matrix of the
Markov chain.
The following Theorem gives the relation between two consecutive state vectors:
If X
n+1
and X
n
are two consecutive state vectors of a Markov chain with transition matrix
T, then
227

X
n+1
=T multiply by X
n Or
X
n+1
=T X
n
We are usually interested in the long-term behavior of a general state vector X
n
in a
Markov Chain. That is we will find the limit of X
n
as n (n approaching infinity).For
regular Markov chains this is always the case. But not so for cyclic chains or for chains
with ergodic sets like the example below.
It may happen that this limit does not exist, for example let

then

As you can see X
n
cycles between the vectors (0, 1) and (1, 0) and therefore does not
reach a fixed vector. Hence T is ergodic, but not regular.
Then what makes X
n
approach a limiting vector as n. This will be explained in next
theorem but first we need a definition:
A transition matrix T is called regular if, for some integer r, all entries of T
r
are strictly
positive. (0 is not strictly positive).
For example, the matrix

is regular since
.
A Markov chain process is called regular if its transition matrix is regular. The main
theorem in Markov chain theory is explain below:
228

1. If T is a regular transition matrix, then as n approaches infinity, T
n
S where S is
a matrix of the form [v, v,,v] with v being a constant vector.
2. If T is a regular transition matrix of a Markov chain process, and if X is any state
vector, then as n approaches infinity, T
n
Xp, where p is a fixed probability
vector (the sum of its entries is 1), all of whose entries are positive.
Consider a Markov chain with a regular transition matrix T, and let S denote the limit of
T
n
as n approaches infinity, then T
n
XSX=p, and therefore the system approaches a fixed
state vector p called the steady-state vector of the system.
Now since T
n+1
=TT
n
and that both T
n+1
and T
n
approach S, we have S=TS. Note that any
column of this matrix equation gives Tp=p. Therefore, the steady-state vector of a regular
Markov chain with transition matrix T is the unique probability vector p satisfying Tp=p.
Is there a way to compute the steady-state vector of a regular Markov chain without using
the limit? Well, if we can solve Tp=p, for p, then yes! You might have seen this sort of
thing before (and certainly will in your first linear algebra course: Ax = b)
Lets me show you an example on how linear algebra can be use to solve p.
If you have lived in London for a while, you must have realized that the weather is a main
concern of the population. An unofficial study of the weather in the city in early spring
yields the following observations:
1. It is almost impossible to have two nice days in a row
2. If we have a nice day, we just as likely to have snow or rain the next day
3. If we have snow or rain, then we have an even chance to have the same the next
day
4. If there is a change from snow or rain, only half of the time is this a change to a
nice day.
a. Write the transition matrix to model this system.
b. If it is nice today, what is the probability of being nice after one week?
c. Find the long time behaviour of the weather.
Answer:
1) since the weather tomorrow depends only on today, this is a Markov chain process.
The transition matrix of this system is
229

where the letters N, R, S represent Nice, Rain, Snow respectively.
2) If it is nice today, then the initial state-vector is

After seven days (one week), the state-vector would be

So, there is about 20% chance of being nice in one week.
3) Notice first that we are dealing with a regular Markov chain since the transition
matrix is regular, so we are sure that the steady-state vector exists. To find it we
solve the homogeneous system (T-I)X=0 which has the following coefficient
matrix:

Reducing to reduced echelon form gives

.

230

The general solution of this system is

So what solution do we choose? Remember that a steady-state vector is in particular a
probability vector; that is the sum of its components is 1: 0.5t+t+t=1 gives t=0.4. Thus,
the steady-state vector is

In the long term, there is 20% chance of getting a nice day, 40% chance of having a rainy
day and 40% chance of having a snowy day.

Conclusion:

In our everyday lives, jobs and business, we make decisions that have consequences. By
deciding to get up in the morning, we have created the possibility that events might occur
in the day. Each of these events has associated probability. Which events occur during the
day will influence what might occur next, and probability is still associated with it. Most
naturally occurring events are not certain. There is always a series of possibilities and
probabilities associated with them. By understanding these, we can understand how
events unfold in our lives, jobs and business, how to optimize our choices, and how to
travel the optimal path with the highest probability. Hopefully with the Markov Chain
examples built with MS Excel above will help you on your day to day decisions making
process.

231

Chapter 7
Bayes Theorem
Introduction

Bayesian Probability

Bayesian probability's application in corporate and business world is highly dependent on
the "degree of belief" rather than historical frequencies of identical or similar events.

You may think that you need to know a lot about probability theory in order to study
Bayesian methods. In fact you don't have to. While any mathematically based topic can
be taken to rather complex depths, the use of a basic Bayesian probability model in
forecasting can help refine probability estimates using an intuitive process.

You can also use your historical beliefs based on frequency to use the model; it's a very
versatile model. Ill show you with 2 examples build with MS Excel on how this is done.

Bayes' Theorem is a theorem of probability theory first proposed by the Reverend
Thomas Bayes. It can be seen as a way of understanding how the probability that a theory
is true is affected by a new piece of evidence.

It has been used in many fields, ranging from marine biology to the development of
"Bayesian" spam blockers for email systems.

It has been used to try to clarify the relationship between theory and evidence in the
philosophy of science. Many insights in the philosophy of science involving
confirmation, falsification, the relation between science and pseudoscience, and other
topics can be made more precise, and sometimes extended or corrected, by using Bayes'
Theorem.

Before we learn more Bayes Theorem, we need to understand the concept of mutually
exclusive

Mutually exclusive
In this section of the tutorial, you will learn the difference between statistically
independent events and mutually exclusive events, sampling with and without
replacement.
Two events are statistically independent if the first event do not affect outcome of the
second event and vice versa. But, these two events may have something in commons
called join events. If the two events do not have something in common, then the two
events are called mutually exclusive events.
232

For example, you have a basket with 5 colourful marbles, 2 marbles has Red colour, 2
marbles has Green colour, and one Blue marble in short [R, R, G, G, B]. Let us do simple
experiments called sampling with replacement and sampling without replacement.
Take two of these marbles without replacement, meaning that after you see the colour of
the first marble you take, record it and immediately take the second marble. Do not return
those marbles into the basket. Now, what is the probability that the first marble you take
is Red and the second marble is Blue? Since you have 5 marbles in the beginning and 2
Red, then the probability that you get Red first marble is 2 out of 5 = 40%. After the first
even, the total marbles in the basket become 4. The probability that you get Blue second
marble is 1 out of 4 = 25%. Thus, the probability that you get Red marble then Blue
marble is 40% times 25% = 10%. (i.e. 0.4 x 0.25 = 0.1)
Now return all marbles in the basket. This time you will take two of these marbles with
replacement, meaning that after you see the colour of the first marble and record it, return
that marble to the basket before you take the second marble. Notice that the total marbles
of the two events are constant, equal to 5, because every time you take a marble, you
return it to the basket before you do the next event. What is the probability that the first
marble you take is Red and the second marble is Blue? You have 2 red marbles out of 5 =
40% and you have 1 Blue marble out 5 = 20%. Multiply these two percentages, we get
40% times 20% = 8%. (i.e. 0.4 x 0.2 = 0.08)
Now the statistical moral story is like this. The two types of sampling have something in
common because they are taken from the same sample of 5 marbles in one basket. In
sampling without replacement, the first event will affect the second event. We say that
the two events are dependent to each other. In the marble experiment, this relation
happens because the total sample change (i.e. reduce) after the first event. The probability
of taking the next event is depending on the previous event. On the other hand, for
sampling with replacement, the first event does not affect the outcome of the second
event. We called the two events are statistically independent. In the marble experiment,
the total sample never changes because you always return the marble that you examine.
The probability of taking the next event is not depending on the previous event. (as
oppose to Markov Chain where the probability of the next event depends only on the current
event and not on the previous event. See Chapter 6)
Bayes' theorem can be stated as follows:
233

Figure 7.1
P(A) is the probability of A occurring, and is called the prior probability.
P(A|B) is the conditional probability of A given that B occurs. This is the posterior
probability due to its variable dependency on B. This assumes that the A is not
independent of B.

P(B|A) is the conditional probability of B given that A occurs.

P(B) is the probability of B occurring.

If we are interested in the probability of an event of which we have prior observations;
we call this the prior probability. We'll deem this event event A, and its probability P(A).
If there is a second event that affects P(A), which we'll call event B, then we want to
know what the probability of A is given B has occurred. In probabilistic notation this is
P(A|B), and is known as posterior probability or revised probability. This is because it
has occurred after original event, hence the post in posterior. This is how Bayes' theorem
uniquely allows us to update our previous beliefs with new information.
Unless you are a world-class statistician, Bayes' theorem (as expressed above) can be
scary. But in reality it is easy to use. The rest of this chapter covers material and
examples that can help you understand when and how to apply Bayes' theorem
effectively.
When to Apply Bayes' Theorem
There are a few factors that we need to know when applying Bayes' theorem. You should
consider Bayes' theorem when the following conditions exist.
234

- The sample space is partitioned into a set of mutually exclusive events { A
1
, A
2
, .
. . , A
n
}.
- Within the sample space, there exists an event B, for which P(B) > 0.
- The analytical goal is to compute a conditional probability of the form: P( A
k
| B
).
- You know at least one of the two sets of probabilities described below.
o P( A
k
B ) for each A
k

o P( A
k
) and P( B | A
k
) for each A
k

Bayes' theorem can be best understood through an example. We will see 2 examples
below on how to apply Bayes' theorem to solve statistical problems.
Example 1: Weather Forecast (Rainy or Sunny)
Susan is celebrating her 30
th
birthday tomorrow. She has planned a huge barbeque party
outdoor in the desert. In the past few years, it has rained only 5 days each year in the
desert. Unfortunately, the weatherman has predicted rain for tomorrow. When it actually
rains, the weatherman correctly forecasts rain 90% of the time. When it doesn't rain, he
incorrectly forecasts rain 10% of the time. What is the probability that it will rain on the
day of Susan's birthday celebration?
Solution: The sample space is defined by two mutually-exclusive events - it rains or it
does not rain. Additionally, a third event occurs when the weatherman predicts rain.
Notation for these events appears below.
- Event A
1
. It rains on Susan's birthday celebration.
- Event A
2
. It does not rain on Susan's birthday celebration
- Event B. The weatherman predicts rain.
In terms of probabilities, we know the following:
- P( A
1
) = 5/365 =0.0136985 [It rains 5 days out of the year.]
- P( A
2
) = 360/365 = 0.9863014 [It does not rain 360 days out of the year.]
- P( B | A
1
) = 0.9 [When it rains, the weatherman predicts rain 90% of the time.]
- P( B | A
2
) = 0.1 [When it does not rain, the weatherman predicts rain 10% of the
time.]
We want to know P( A
1
| B ), the probability it will rain on the day of Susan's birthday
celebration, given a forecast for rain by the weatherman. The answer can be determined
from Bayes' theorem, as shown below.

235

Figure 7.2
Note the somewhat unintuitive result. Even when the weatherman predicts rain, it only
rains only about 11% of the time. Despite the weatherman's gloomy prediction, there is a
good chance that Susan will not get rained tomorrow on her birthday celebration.
Example 2: Stock Market Index
For this example, we will be using the rules and assertions of the school of thought that
pertains to frequency rather than subjectivity within Bayesian probability.

This means that the measurement of knowledge that is being quantified is based on
historical data. This view of the model is where it becomes particularly helpful in
financial modelling. The application of how we can integrate this into our models is
explained in the section to follow. Lets see how it works while incorporating Bayes'
theorem within an equity market concept.

Let's say we want to know how a change in interest rates would affect the value of a stock
market index. All major stock market indexes have a plethora of historical data available
so you should have no problem finding the outcomes for these events with a little bit of
research. For our example we will use the data below to find out how a stock market
index will react to a rise in interest rates.

Here:

P(SI) = the probability of the stock index increasing
P(SD) = the probability of the stock index decreasing
P(ID) = the probability of interest rates decreasing
P(II) = the probability of interest rates increasing

So the equation will be:

Thus with our example plugging in our number we get:

In the table you can see that out of 2000 observations, 1100 instances showed the stock
index decreased. This is the prior probability based on historical data, which in this
example is 55% (1100/2000). This probability doesn't take into account any information
about interest rates, and is the one we wish to update. After updating this prior probability
236

with information that interest rates have risen leads us to update the probability of the
stock market decreasing from 55% to 80%. 80% is the posterior probability.

Open the file Bayes.xls in the Chapter 7 folder. Select worksheet Stock Price.

Figure 7.3

Modelling with Bayes' Theorem
As seen above we can use the outcomes of historical data to base our beliefs on from
which we can derive new updated probabilities. This example can be extrapolated to
individual companies given changes within their own balance sheets, bonds given
changes in credit rating, and many other examples.

So what if one does not know the exact probabilities but has only estimates? This is
where the subjectivists' view comes strongly into play. Many people put a lot of faith into
the estimates and simplified probabilities given by experts in their field; this also gives us
the great ability to confidently produce new estimates for new and more complicated
questions introduced by those inevitable roadblocks in financial forecasting. Instead of
guessing or using simple probability trees to overcome these road blocks, we can now use
Bayes' Theorem if we possess the right information with which to start.

Now that we have learned how to correctly compute Bayes' Theorem, we can now learn
just where it can be applied in financial modelling. Other, and much more inherently
complicated business specific, full-scale examples will not be provided, but situations of
where and how to use Bayes' Theorem will.

Changing interest rates can heavily affect the value of particular assets. The changing
value of assets can therefore greatly affect the value of particular profitability and
efficiency ratios used to proxy a company's performance. Estimated probabilities are
widely found relating to systematic changes in interest rates and can therefore be used
effectively in Bayes' Theorem.

Another avenue where we can apply our newfound process is in a company's net income
stream. Lawsuits, changes in the prices of raw materials, and many other things can
heavily influence the value of a company's net income. By using probability estimates
relating to these factors, we can apply Bayes' Theorem to figure out what is important to
us.
237

Once we find the deduced probabilities that we are looking for it is only a simple
application of mathematical expectancy and result forecasting in order to monetarily
quantify our probabilities.

Conclusion

And that's Bayes' Theorem. Rational inference on the left end, physical causality on the
right end; an equation with mind on one side and reality on the other. Remember how the
scientific method turned out to be a special case of Bayes' Theorem? If you wanted to
put it poetically, you could say that Bayes' Theorem binds reasoning into the physical
universe.
To conclude, we found that by using a myriad of related probabilities we can deduce the
answer to rather complex questions with one simple formula.

238

General Conclusion
After reading the book, you can start to create accurate forecasts quickly and easily using
proven statistical forecasting methods. Research has shown that no single method
works best for all data, which is why Ive provides a comprehensive range of popular
forecasting approaches to address all types of business and individual needs.
I have also given an overview of the types of forecasting methods available. The key in
forecasting nowadays is to understand the different forecasting methods and their relative
merits and so be able to choose which method to apply in a particular situation. All
forecasting methods involve tedious repetitive calculations and so are ideally suited to be
done by a computer. Using MS Excel as a powerful forecasting tool is a good starting
point before you delve into expensive forecasting software. The user's application and
preference will decide the selection of the appropriate technique. It is beyond the realm
and intention of the author of this book to cover all these methods. Only popular and
effective methods are presented.
You can use this book spreadsheets models as reference to build statistically-based
forecasts that is powerful yet remarkably easy to use. A full range of custom modeling
spreadsheets and detailed diagnostic tools are included to support even the most
sophisticated analysis. And you can do much more after reading this book, including
adding your business knowledge, creating dazzling presentations and working with your
existing data.

239

Appendix A

This book comes with an Excel add in nn_Solve and many mathematical models
developed with Excel spreadsheets.

Installing the nn_Solve add-in :

1. Open the folder nn_Solve.zip. You will see the addin nn_Solve. Double click on
it. This will launch nn_Solve.

2. Select Enable Macros (see below)

3. You will see nn_Solve on the Excel menu bar like below:

4. If you cant open the addin nn_Solve, then from the Excel menu bar, select Tools
-> Options -> Security tab. Select Macro Security on the bottom right and Select
Medium. (see below)
240

5. Double click on nn_Solve icon again. This will open nn_Solve
241

Appendix B
Simple Linear Regression For Forecasting
Excel 2007 has a built in regression analysis tool, packaged as part of its Analysis
Toolpak. The Analysis ToolPak has a comprehensive approach to simple linear
regression and returns much useful statistical in regards to the subjected data.

You can add Analysis Toolpack with these steps:
Excel options > Add-Ins > Go > Select Data Analysis Toolpack & Toolpack VBA. Data
Analysis now available under Excels Data tab

In linear regression, you are seeking to describe the association between two data sets in terms
of the equation for a straight line: y = a + bx.
In this part of the book, we are going to
a) How to carry out a Model I simple linear regression: drawing a regression line
b) How to carry out a Model I simple linear regression: testing the significance of
the association
Step 1
Open a new worksheet in Excel 2007 and enter the data shown in Figure A1.2
Enter the data sets in columns using appropriate labels in the first row.

242

Step 2.

Select and click on the Data tab.
Ensure that Analysis ToolPak has been added to the Data Analysis tool using Add-
ins as described above. Click on Data Analysis to open the Data Analysis box. From
the Analysis Tools list, select Regression. Click on OK.

Step 3. In the Regression dialogue box that opens, enter the input variables.

You must determine which set of data is to be plotted on the x-axis (the independent
variable) and which on the y-axis (the dependent variable). In this example, distance
from the pylon is the independent variable and so the cell locations for these data,
including the data label, must be entered into the box alongside Input X Range. Click in
this box to ensure that the cursor is flashing there and then click on cell A1 and, holding
243

down the left mouse button, drag down to cell A6 and let go. The cell locations will now
be entered into the box. Repeat this procedure to enter the dependent data set
into the Input Y Range box. As the data labels have been included in the data range,
click on the Labels box so the tick appears. (If this is not done, Excel will try to read the
fi rst row as numerical data and an error will occur.)
There is an option to select Constant is Zero. This will assign the value of zero to the
constant term a in the equation y = a + bx and force the regression line through the origin
(point 0,0). For some data sets, this property is very useful, but it is not appropriate in this
example, as we anticipate a negative association between these data sets.
In the Input box, you can also select the confi dence level and this defaults to 95% so
normally would not need to be changed.

Step 4.

Outputting the data.

There are a number of output options. First, select where your results will be returned. If
you select Output Range, remember to click on the radio button and then also to click in
the adjacent box to bring the cursor to that location (otherwise the input data will be
changed). Click on a square or drag across an area to select where the results will be
returned. Excel will determine the number of cells it requires for the output so you do not
have to. Alternatively, you can select to have the output on a New worksheet or a New
workbook by clicking the appropriate radio buttons.

Under Residuals, you can select all items by clicking in each box:
244

Click on Residuals to include residual values in the residuals output table. Click on
Standardized Residuals to include standardized residual values in the residuals output
table. Click on Residual Plots for a chart (automatically generated) showing each
independent variable versus the residual value. Click on Line Fit Plots for a chart
(automatically generated) showing predicted values versus the observed values. Having
selected all the desired options, click on OK.

Step 5.

Interpreting the data.

The first box gives a summary output and returns values for R (the product moment
correlation coeffi cient) and R2 (the coeffi cient of determination). Note that both these
values are approaching 1.0 and so there is an extremely strong correlation between the
two variables. Adjusted R2 gives an adjusted value in relation to the number of
observations.
An ANOVA table is given. The ANOVA tests the signifi cance of the regression line. A
value for F is given (96.96) and its signifi cance (p = 0.002) and therefore the regression
line is statistically significant.
In the following table, Intercept gives the value for a in the equation: y = a + bx. Here a
= 737.6. Distance from the pylon gives the value of b in the equation and is the
gradient of the regression line (i.e. the gradient of the slope). Note that the value is
negative, as one of the variables increases as the other decreases (b = 92.4).
For each coeffi cient, values are assigned for the standard error. t statistics are also
generated, which test the levels of signifi cance of the coefficients. The probabilities for
245

these statistics are given; note that both are signifi cant (p < 0.005 for the gradient (actual
value p = 0.002) and p < 0.00005 for the intercept value, p = 4.32 105).
Residual values are given: these indicate the distance of the actual data points from the
regression line. These values can be plotted above and below a horizontal line
representing the regression line inthe residual plot chart, which facilitates a visualization
of the spread of values around the line. The charts are returned stacked. It is necessary
to drag them apart and edit the charts as required particularly expanding the plot area
(see below for line fit plot).

Step 6.

Presenting data.

Click on the top chart to select it and drag it to an empty part of the worksheet. Do the
same to the second chart. Now expand and edit the charts to improve their legibility and
appearance. Having expanded the line fi t plot, you can see the actual values plotted as
blue diamonds and the predicted values as red squares. The predicted values give the
position of the regression line calculated from the data. In this chart, the two sets of
values almost coincide.
In order to add the regression line, move the cursor over one of the red squares and click
with the right mouse button. This highlights all the red squares and they appear to have a
blue cross behind them. Simultaneously, a pop-up menu appears.

246

From the menu, click on Add Trendline and a new box will open. Here you can select
the type of trend line required and also format it.

247

Here we will choose the linear trendline and also tick the boxes for Display Equation
on chart and Display R-squared on chart. You can also select trendline options for
colour, style, and shadow on the trendline.
The Forecast box enables the fitted regression line to be extended either side of the
regression line. This is a useful facility in some cases, but note that, on most occasions
with biological and environmental data, this must not be done, as it would be erroneous to
assume values beyond the data points. The next option is to set the intercept to zero or
another chosen value. This is not required in this example.
To remove the redundant lower values on the y-axis, click on the y-axis to highlight it and
make sure the formatting boxes are open. Click on the attributes you want to change and
select appropriate changes from the dialogue boxes.
To add grid lines, ensure that the chart is selected by clicking on it. The Chart Tools
should now become available on the tool bar. Click on the Layout tab and then
Gridlines from the Axes section.
248

249

Regression line Equation and R2 values

250

The Residuals plot can also be expanded in the same way as the line fit plot. Here the
relationship of the actual data points to the regression line is shown. The regression line is
shown horizontally and the data points lie above and below this line where the values are
greater or less than the regression line, respectively. The greater the deviation of the
points from the regression line, the smaller would be the value for R2.
The chart above shows the residual plot of the association between zinc concentration in
the soil and distance from the pylon (m).
251

Appendix C:
Accessing Excel Solver 2007
The Solver Add-in is a Microsoft Office Excel add-in program that is accessible when
you install Microsoft Office or Excel. To use it in Excel, however, you need to load it
first.

Follow the instructions below to load the Solver Add-in:
1. Click the Microsoft Office Button, and then select Excel Options from the bottom
right-hand corner of the box.

2. Switch to the Add-Ins tab, and then in the Manage box, select Excel Add-ins. Then
click Go.
252

3. In the Add-Ins available box, select the Solver Add-in check box, and then click
OK.

NOTE: If Solver Add-in is not listed in the Add-Ins available box, click Browse to
locate the add-in. If you get prompted that the Solver Add-in is not currently installed on
your computer, click Yes to install it.

4. After you load the Solver Add-in, the Solver command is available in the Analysis
group on the Data tab. (right hand side)

Powerful Forecasting With MS Excel Sample

Uploaded by

Copyright:

Available Formats

Powerful Forecasting With MS Excel Sample

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Powerful Forecasting With MS Excel Sample

Uploaded by

Copyright:

Available Formats

What are some of the forecasting methods discussed in the book?

What chapters are covered in the book?

Readers will be provided a link to download the software

You might also like