SEEM 4610 Forecasting

Demand Forecasting in
Supply Chain

Prepared & presented by:

Sharfuddin Lisan
BBA,MHRM(DU),SCEM, PGDSCM, MBA(SCM)-Canada

[email protected], [email protected]
01731822888

Learning Objectives
 Role of forecasting in a supply chain
 Components of a demand forecast
 Demand forecasting using historical data
 Analysing forecast errors
 Managing demand/supply in a supply chain
 Forecasting with Excel

BIHRM Supply chain

Janny Leung 1
SEEM 4610 Forecasting

Role of forecasting
 Demand forecasts form the basis of all
planning decisions in a supply chain
 Push: produce to anticipated demand levels
 Pull: set capacity and component availability
levels
 Forecast time horizons
 Short term (days, weeks): shift scheduling
 Medium Term (weeks, months): workforce
planning, materials purchasing, promotions
 Long term (months, years): capacity expansion,
capital/financial budget
BIHRM Supply chain

Characteristics of Forecasts
 Forecasts are always wrong!
 Expected value
 Error/variability from the expected value
 Long-term forecasts are usually less accurate than short-term
forecasts
 Aggregate forecasts are usually more accurate than
disaggregate forecasts
 Mature products with stable demand are easier to forecast than
seasonal goods or “fashion” items with short product-life

BIHRM Supply chain

Janny Leung 2
SEEM 4610 Forecasting

Influences on Customer Demand

“Predictions are usually difficult, especially about the
future” – Yogi Berra

 Historical patterns
 Past demand -> future demand
 Seasonality? Trend?
 Externalities
 Weather
 State of the economy
 Internal factors
 Planned promotional/discount campaigns
 Display position and advertising efforts
 Competitors’ actions
BIHRM Supply chain

Components of Observed
Demand Observed demand (O) =
Systematic component (S) + Random
component (R)
 Forecasting should focus on identifying the
systematic component
 Systematic component = Expected value of demand
 Time series model:
 (Basic) demand level
 Trend, rate of growth/decline in demand per
period
 Seasonality, (predictable) seasonal fluctuations
BIHRM Supply chain

Janny Leung 3
SEEM 4610 Forecasting

Forecasting Methods
 Qualitative
 Subjective, human judgement and opinions
 Little historical data available, e.g. new product, .com
 Time Series
 Assume past history is good indicator of future demand
 Best for stable environments
 Easy to implement
 Causal
 Assume other (measurable) factors that is correlated with
demand
 Models (e.g. regression) identify factors and quantifies the
strength of the correlations
 Simulation
 Assumes some underlying principles of customer behaviour
and develop possible scenarios in the future to predict
demand
BIHRM Supply chain

Basic Approach to Demand Forecasting

 Understand the objective of forecasting
 Forecast horizon; affected by suppliers’ lead times
 Integrate demand planning and forecasting
 Co-ordination between marketing, production and suppliers
 Identify major factors that influence the demand forecast
 Growth Trend? Seasonality?
 Substitutes and complementary products?
 Understand and identify customer segments
 Levels of aggregation
 Determine the appropriate forecasting technique
 Geographical location, product life-cycle, etc.
 Establish performance and error measures for forecasts
 Assess cost impacts; consider investments in improving forecasts
accuracies

No forecast is perfect; system must be flexible and have contingency

plans to handle forecast errors

BIHRM Supply chain

Janny Leung 4
SEEM 4610 Forecasting

Common Time Series

Patterns
 Constant (stationary)
 Increasing/decreasing linear trend
 Seasonal variations
 Non-linear (e.g. exponential) trend
 Combinations
 Additive
 Multiplicative
 Mixed

BIHRM Supply chain

Common Time Series Patterns

Purely Random Error - Increasing
No Recognizable Pattern Linear Trend
Demand
Demand

Time Time

Seasonal Pattern Seasonal Pattern plus

Demand

Linear Growth
Demand

BIHRM Supply chain

Time Time

Janny Leung 5
SEEM 4610 Forecasting

Other Time Series Patterns

Demand Curvilinear Trend

(quadratic, exponential)

Time
Change of variables?
BIHRM Supply chain

Special Forecasting
Difficulties for Supply Chains
 New products and service introductions
 No past history
 Use qualitative methods until sufficient data collected
 Examine correlation with similar products
 Use a large exponential smoothing constant
 Lumpy derived demand
 Large but infrequent orders
 Random variations “swamps” trend and seasonality
 Identify reason for lumpiness and modify forecasts
 Spatial variations in demand
 Separate forecast vs. allocation of total forecasts

BIHRM Supply chain

Janny Leung 6
SEEM 4610 Forecasting

Factors affecting promotion timing

Factor Favoured promotion time
High forward buying Low demand period

High stealing market share High demand period

High market growth High demand period

High margin product High demand period

High inventory holding costs Low demand period

Low production flexibility Low demand period

BIHRM Supply chain

Demand management
 Changes in demand impact production
scheduling and inventory levels
 Promotion at peak demand periods increases
demand variability
 Promotion at non-peak “smoothes” demand
 Pricing and production planning must be
done jointly
 Preempt, don’t just react, to predictable
variability
 Actively managing predictable variability
can be a strategic competitive advantage
BIHRM Supply chain

Janny Leung 7
SEEM 4610 Forecasting

Flexibility and Quick Response

No forecast is perfect …
There is no better forecast than the actual order!
 If supply chain is responsive, then effect of
forecast errors is minimised
 Especially important when demand is
unpredictable
 Example: National bicycle
 Difficult to predict styles in sports bikes
 Re-engineer supply chain
 Customers design their own bike on Internet
 Bike produced
BIHRM Supply chain
in Kashiwara and delivered in 2 weeks!

De-seasionalized
linier regression
Seasonalized
Forecaste

Year Quarter Period, t Demand

d,
1 8,000
D t
2 13,000
3 23,000
4 34,000
5 10,000
6 18,000
7 23,000
8 38,000
9 12,000
10 13,000
11 32,000
12 41,000

Table 7-1
BIHRM Supply chain

Janny Leung 8
SEEM 4610 Forecasting

Paradise Salt

Figure 7-1

BIHRM Supply chain

Estimate Level and

Trend
Periodicity p = 4, t = 3
t –1( p /2)

Dt –( p /2)  Dt( p /2)  2Di / (2 p) for p even

it1–( p /2)
Dt 
t[( p –1)/2]

Di / p for p odd
it –[( p –1)/2]

t –1( p/2)
Dt  Dt –( p/2)  Dt( p/2)  2Di / (2 p)
it1–( p/2)
4
D D  2D / 8
1 5 i
i2

BIHRM Supply chain

Janny Leung 9
SEEM 4610 Forecasting

B+ PS +Ending
6000+2(1300+2300+3400) +1000/8  xy  n x y
b 2 a  y  bx
Y11=a+bx  x2  n x
X Y XY X sqr Seasonalized SI
1 6000 2387.854+-36.428x1 2.59
=2351.426
2 1300 2314.998 0.56
3 2300 2625 7875 9 2278.57 1.0094
4 3400 1837.5 7350 16 2242.142 1.516
5 1000 2125 10625 25 2205.714 0.453
6 1800 2175 13050 36 2169.286 0.829
7 2300 2250 15750 49 2132.858 1.078
8 3800 2112.5 16900 64 2096.43 1.812
9 1200 2060.002 0.582
10 1300 2023.574 0.642
71550-6x5.5x 2187.5
11 2073.98 1987.146 x 1.043
b=------------------------- = -36.428
199-6x30.25
BIHRM Supply chain

a= 2187.5-(-36.428)x5.5=2387.854

Paradise Salt

Figure 7-3

A linear relationship exists between the

deseasonalized
demand and time based on the change in demand over time
D LT
t t
BIHRM Supply chain

Janny Leung 10
SEEM 4610 Forecasting

Linier Regression

 xy  n x y
x2
b 2
x y xy xy x2
 x2  nx
3 19750 59250 9
4 20625 82500 16
a  y  bx
5 21250 106250 25
6 21750 130500 36
7 22500 157500 49 X avg 6.5
Yavg 21843.75
8 22125 177000 64
22000
9 22625 203625 81
42
10 24125 241250 100 b= 523.8095
a= 18438.99
SUM 52 174750 1157875 380
y=a+bx
BIHRM Supply chain
18435+524 x

Estimating Seasonal Factors

Di
St 
Dt

Figure 7-4
BIHRM Supply chain

Janny Leung 11
SEEM 4610 Forecasting

Estimating Seasonal Factors

r –1
S jp1
Si  j0

r
S  (S  S  S ) / 3  (0.42  0.47  0.52) / 3  0.47
1 1 5 9

S 2  (S 2  S 6  S10 ) / 3  (0.67  0.83  0.55) / 3  0.68

S 3  (S 3  S 7  S 11 ) / 3  (1.15  1.04  1.32) / 3  1.17
S 4  (S 4  S 8  S12 ) / 3  (1.66  1.68  1.66) / 3  1.67

F  ( L  13T )S  (18,439  13 524)0.47  11,868

13 13

F14  ( L  14T )S 14  (18,439  14 524)0.68  17,527

F15  ( L  15T )S 15  (18,439  15 524)1.17  30,770
F16  ( L  16T )S 16  (18,439  16 524)1.67  44,794
BIHRM Supply chain

Thanks

BIHRM Supply chain

Janny Leung 12