1

PRODUCTION
PLANNING AND
CONTROL
 UNIT - 2 2

SALES FORECASTING
 A forecast is an estimate of the level demand to the expected for a production of
several products for some period of time in the future.
 Forecast is made of sales ( in Rs.) or physical units under a proposed marketing or
program & under an assumed set of economic & other force outside the unit
system).
 Forecast should cover a time period at least as long as the period of time required
to make the decision & to put that decision into effect.
 3
Applications/uses/purposes :
There are 3 major purposes.
1) To determine necessity for & the size of plant expansion ( Facility Forecast )
2) To determine intermediate planning for existing products to be manufactured with
existing facilities.
3) To determine Short-time scheduling of existing products to be manufactured on
existing equipment ( Product Forecast )
 Forecasting for new product
4
 1. Direct Survey method.
 2. Indirect Survey method.
 3. Comparing with established product.
 4. Limited market trial.

Forecasting for a new product is difficult task as no past information is available to

predict the future.
Forecasting for established product 5
 1. Related information method.
 2. Market Research.
 3.Sales force composite method or Subjective opinion method.
 4. Jury of Executive opinion.
 5. Projection method.
 6. Statistical method.
1) Direct Survey method : 6
In this method, representative sample of customers are approached & asked, what
they intend to buy. By doing so, it is possible to predict, with some degree of
certainly how the population will respond. Here economic, political, changes,
customs, habits, social requirements are considered.
2) Indirect Survey method :
In this method the attitude & behavior of the customers is predicted through
salesman, agents, whole sellers, retailers etc.
3) Comparing with Established Product :- At times the product under consideration is
comparable to an existing product. So sales figures can be compared.
4) Limited Market Trial : Some times limited selling technique is adopted to product
acceptance of the product.
Forecasting for Established product is explained below. 7
1) Related information method : ( Forecast based on an index ) in this method
which directly varies with the sales volume is found (e.g. birth date is related
information for the forecast of the sales of baby food & forecast is made ).
2) Market Research : Through critical analysis of the marketing forces, changing
pattern of socio-economics pressures, political changes in style, attitudes, fashion
etc, we can diet the future demands of the product.
3) Sales force composite method : ( Subjective opinion ) in this forecast, all the
marketing & sales peoples ( Sales-man, traders middle men etc ) express their
considered opinion of the volume of the sales expected in the future. These
opinions are then collected & evaluated.
4) Jury of Executive Opinion Method : Here opinions of experts are invited about the 8
sale in future. It is simple & fast but not scientific.

5) Projection Method : Based on the historical data, future can be projected to some
extension. A line drawn through known information is projected into the forecast area
to predict what the sales volume will be for future periods. Projection of future can be
done either by time series Analysis or Correlation, regression Analysis tech.
 9

 Time series Analysis : A time series is a chronological data which has some
quantity such sales rupees, sales volume, prod. figures, imports, exports, number
of passengers traveling by Air lines etc. as dependent variable & unit of time as
independent variable.
 The movements or variations of the dependent variable may be Long period
changes (Trend)
 Short period changes ( Seasonal, cyclic, irregular )
 10

 Long period tendency of the data to increase or decrease is called secular or basic
trend. A secular or long term trend refers to the smooth & regular movement of a
series reflecting continuous growth, stagnation or decline over a rather long
period of time..
 We are chiefly concerned with the general tendency of data. As long as we notice
an upward or down word trend movement in the data over the whole period, we
conclude secular trend.
 e.g. There is secular rise in agriculture production in India. Because of since last
25 year it has been found that except for a year or two, the production is
increasing. Basic secular trend ( long term ) may be linear or non-linear.
 11

 The important method of making inference about the future on the basis of what
happened in the past is the analysis of time series. After analysing the time series
plant production, finance, personnel, marketing etc. can be made to meet the
variation in the demand.
 Time series is a multiplicative model of 4 – components.
 O= T x S x C x I
 T - Basic Trend, S – Seasonal fluctuations, C - Cyclic fluctuations,
1 – irregular, Erratic, random, fluctuations.
METHODS OF ESTIMATING TRENDS : 12
 1. Methods of Inspection or freehand methods – Once the given time series data
have been plotted on a graph paper, a line is drawn through the points which in
statisticians opinion best describes the avg. long term growth.
 2. Methods of Averages :
 a) Selected point method : Here the values are selected of the years which are
considered to be the most representative or normal. Then a straight line is drawn.
 b) Semi-avg. method : Here data is divided into two equal halves & averages for
each half is calculated. The avg. for half is taken to be representative of the value
corresponding to the mid-point of the time interval of that half. Thus these two
points determine the position of a trend line. The trend line can be extended to
estimate future values or intermediate values.
 c) Moving Avg. methods :
 3) Statistical methods : It is statistical analysis of past demand. It is most accurate
method provided there is relationship between past & future. In fact past offers 13
best basis for decision on future action. However one must modify the prediction
form past data if he knows that certain events will or will probably happen in
future e.g. events like expansion of sales area, advertising, withdrawal of
competitor etc., will tend to increase the sale.
On the other hand events like entry of new competition, product
becoming calculated will tend to develop sale. Such considerations should be
reflected in new forecast.
In statistical method we are going to consider following 3 cases.
1. Level demand with Random variation.
2.An Upward Trend with Random Variation.
3. Cyclic (Seasonal) Demand.
 General Approach To Statistical Forecasting : 14
 If we assume that use of statistical methods applied to past data is a realistic of
forecasting future demands, we should then proceed as follows.
 1. Make a plot of demand versus time. ( Demand as ordinate & time as abscissa).
 2. Determine which statistical technique to try.
 3. Evaluate the expected error.
 4. Make a decision to use the technique under consideration or attempt to find a
being one.
 Above approach is demonstrated by solving three examples below.
 15

 Example 1. Level Demand With Random Variation. In this case demand remains
essentially constant but super imposed random variation. The use of constant
forecaster is generally adequate & appropriate.
 16

 Constant forecaster is given by –

 d = total demand / number of period
=1191 / 12
= 99.25
99 Units can be used as a forecasting function. That is we would forecast that demand
will be 99 units for each of the next several months.
 17

 We want to evaluate this method (arithmetic avg.) forecasting . This can be done
by determining the standard error of estimation or standard deviation.
 s= square root of (∑sauare(d-D)/n-1)
 D= actual demand
 n - No of periods included
 d - Avg. demand
 The calculation are shown in tabular form.
18
 s=square root of [1180/ (12-1)]=10.4 19
 Now we can say that we are 95% confident that the demand in any month will lie
between 79 and 119 units. 95% confidents means, in 95 month out of next 100
months demand will lie between 79 and 119.
 20

 Assumptions made:-
 1. Demand data studied for the periods are truly representative of the demand.
 2. The cause system was unchanged & unchanging during the period studied.
 3. Same cause system will continue to be operative for some time into the future.
21
Method of Moving Averages 22
 A moving average may be defined as an average of some fixed or predetermined
number of observations in a time series, which moves through the series by
dropping the top item of the previous averaged group and adding the next item
below in each successive average.
 The calculations depend upon the period to be odd or even. In case of odd periods
generally 3,5,7 or 9 years periods are taken to compute moving averages.
 The method of calculating moving average is simple as explained.
 When we calculate 5 yearly moving average the first five year’s values are added
and their total and average will be recorded against the third year(i.e. against the
center of the period used in the calculation of average).
 Next we eliminate the first year’s value and add the next five years’ values and
note the total and then mean against the fourth year. 23
 With a three period moving average, there can be no average against the first and
last time periods. With a five years average there can be no average for the first
two and last two time periods. With a seven years average there can be no average
for the first three and last three time periods.
 If the period of observations is even like 4 or 6 years etc., then the average of the
four yearly observations is written between second and third year values. After
this centering is done by finding the average of the paired values. The even order
periods create the problem of centering between the periods and therefore
generally odd order periods are preferred.
 The calculated values of moving averages become the basis for determining the
expected future sales.
Advantages:
24
 The moving average method provides a simple and good estimate. In this method
equal weightage is assigned to all the periods chosen for averaging.
 The process of averaging lessens influence of the fluctuations.
 It is more accurate than graphical method as it is based on mathematical
calculations.
Disadvantages:
 Records of the demand data have to be maintained for a fairly long period.
 If demand series depicts trends as against the stationary level, the moving average
method would provide forecasts that lags the original series.
 Choice of period of moving average is difficult.
 Cannot be applied if some observations are missing.
25
26
27
28
Exponential smoothing method 29
 It is similar to moving averages and used fairly extensively and is improvement
over the method of moving average of forecasting.
 It tries to overcome the limitations of moving averages and eliminates the
necessity of keeping extensive records of past data.
 It also tries to screen out the irregularities in demand pattern.
 It represents a weightage average of the past observations.
 The most recent observation is assigned to the highest weightage and it decreases
in geometric as we move towards older observations.
 This method is particularly useful when forecasts of a large number of items are
made.
 This method is also adaptable for trend corrections and smoothing of forecast
errors.
 It is one of the most accurate statistical techniques for forecasting, because the 30
most recent observations are given more weightage which are likely to reflect
more up to date average.
 The fundamental concept of exponential smoothing method is that
new estimate= old estimate of latest actual demand+( latest actual demand –old
estimate of latest actual demand)

(or) = + et
where et = -
Ft is the forecast at time t
Dt is the actual demand at time t
Ft-1 is the forecast at time t-1
is the smoothing coefficient
 The procedure is summarized as below
 (i) Find error by subtracting the recent average from the latest incoming 31
observation.
 (ii) Multiply error et with . This is the correction to be applied to the past average.
 (iii) Add correlation et to the past average . This gives new average as the
forecast for the next period.
 The performance of this method depends on the value of the smoothing
coefficient and initial forecast .
 The selection of value for depends on how much weightage is desired to be given
to later periods relative to earlier periods.
 A low value of gives more weightage to the past figures and less consideration to
incoming observation.
 Low values of the smoothing coefficient are used where the series is rather stable
and high values where the series is fluctuating. The values of lies between 0 and
1.
 If cyclic fluctuations are predominant in forecast then low value of should be 32
selected; and if long term fluctuations are more predominant then select high
value of .
 Two main factors should be considered in selection of
 (i) Distribution of random errors and
 (ii) Cost associated with forecasting errors.
 In practice is generally chosen between 0.1 and 0.3.
 The choice of initial forecast is either based on subjective estimates or on simple
arithmetic average of the past few periods.
 Exponential smoothing provides a convenient, systematic, and recursive method
for revising the forecast for the next period whenever discrepancy exists between
the previously forecast demand for the current period and actual demand for
current period.
33
34
 35

 Solution: From the graph, the demand for the year comes out to be approximately
25000 units.
Least Square Method 36
 Many methods can be used to fit a straight line to a given scatter which suggests a
linear trend but the most widely accepted method is that of least squares.
 By using this, a straight line is defined in such a way that the sum of the squares
of the differences between the ordinates of the suggested line and those at given
points is at a minimum.
 A mathematical relationship is established between the time factor X and variable
Y.
 Let Y denote demand and X denote period for a certain product.
 Then Y and X are related as Y= a + bX
 Here “a” is Y intercept and “b” is slope.
37
 38

 We can find out the variable X from any point of time series taken as origin such
as the first one. But to simplify the calculations the mid point of time series is
taken as the origin.
 In this process, the negative values in the half of the series balance out the
positive values in the other half.
39
 40
 While using the method of least squares to compute a trend line, it is convenient
to use the middle of the time series as origin.
 If the series consists of an odd number of years, the origin is the middle year.
 If an even number of years is given, the origin falls between the middle two years.
Calculations are slightly different in the case of even number of years.
41
42
43
 44

Advantages:
 There is no need to conduct any sample survey as only past information about
sales is required.
 Method is simple and easy to understand
 Under normal situations method gives accurate results.
Disadvantages:
 The method is based on mathematical formulae which may not be understood by
common people.
 The assumptions that other things remain constant may not hold good in practice.
45
Qualitative methods of forecasting 46
 Qualitative forecasting methods, often called judgmental methods, are methods in
which the forecast is made subjectively by the forecaster.
 They are educated guesses by forecasters or experts based on intuition,
knowledge, and experience.
 When you decide, based on your intuition, that a particular team is going to win a
baseball game, you are making a qualitative forecast. Because qualitative methods
are made by people, they are often biased. These biases can be related to personal
motivation, mood, or conviction.
47
Principles of forecasting 48

 At any point in time, there is only one best estimate forecast for a project that
reflects the current understanding of subsurface uncertainty and best development
and commercial assumptions.
 This forecast should always be accompanied by an uncertainty range.
 The forecast uncertainty range should always have remaining reserves as an
objective function.
 The forecast and uncertainty range should be based on defined projects, with
incremental forecasts for subsequent projects.
 49
 Quantitative methods: These types of forecasting methods are based on
mathematical (quantitative) models, and are objective in nature.
 They rely heavily on mathematical computations.
 Time-Series Models
 Time series models look at past patterns of data and attempt to predict the future
based upon the underlying patterns contained within those data.
 Associative Models
 Associative models (often called causal models) assume that the variable being
forecasted is related to other variables in the environment.
50
51
 DECOMPOSITION OF A TIME SERIES 52
 Patterns that may be present in a time series
 Trend: Data exhibit a steady growth or decline over time.
 Seasonality: Data exhibit upward and downward swings in a short to intermediate
time frame (most notably during a year).
 Cycles: Data exhibit upward and downward swings in over a very long time
frame.
 Random variations: Erratic and unpredictable variation in the data over time with
no discernable pattern.
 Dependent versus Independent Demand
53
 Demand of an item is termed as independent when it remains unaffected by the
demand for any other item. On the other hand, when the demand of one item is
linked to the demand for another item, demand is termed as dependent.
 It is important to mention that only independent demand needs forecasting.
Dependent demand can be derived from the demand of independent item to which
it is linked.
 Business Time Series
 The first step in making a forecast consists of gathering information from the 54
past. One should collect statistical data recorded at successive intervals of time.
Such a data is usually referred to as time series.
 Analysts plot demand data on a time scale, study the plot and look for consistent
shapes and patterns.
 A time series of demand may have constant, trend, or seasonal pattern or some
combination of these patterns. The forecaster tries to understand the reasons for
such changes, such as, changes that have occurred as a result of general tendency
of the data to increase or decrease, known as secular movements.
 Changes that have taken place during a period of 12 months as a result in changes
in climate, weather conditions, festivals etc. are called as seasonal changes.
 Changes that have taken place as a result of booms and depressions are called as
cyclical variations.
 Changes that have taken place as a result of such forces that could not be
predicted (like flood, earthquake etc.) are called as irregular or erratic variations.
55