Forecasting
Forecasting
Forecasting
What is Forecasting?
• Process of predicting a
future event on the basis of Sales will
past as well as present be $200
knowledge and experience Million!
• Underlying basis of
all business decisions
– Production
– Inventory
– Personnel
– Facilities
• To reduce risk and uncertainty
• For planning
Steps involved in Demand Forecasting
• Information required
– Disposable income of consumer
– Price of the product
– Price of related products
– Demography
Selection of Method
• Time horizon
• Stability
• Data Availability
• Cost
• Accuracy
• Ease of Application
Qualitative Forecasts & Sources of Data
Complete Sample
enumeration survey
Delphi Method
• Iterative group
process Decision Makers
• 3 types of people (Sales?)
Staff (Sales will be 50!)
– Decision makers
– Staff (What
will sales
– Experts be?
• Proposes to reduce survey)
‘group-think’ Experts
(Sales will be 45, 50, 55)
Demerits
• Expensive
• Time consuming
• Biased opinion
Consumer Market Survey
• Ask customers How many hours will
about purchasing you use the Internet
plans next week?
• What consumers
say, and what they
actually do are
often different
• Sometimes
difficult to answer
Merits & Demerits
• Insufficient Information
• Lack of time
• Biased information
• Utility limited to very short period
• There may be sampling error, if sample is not properly
chosen
Sales Force Composite
• Simple
• Based on first-hand knowledge of salesman
• Biased opinion
• Restricted to short-term forecasting
Market Experiments
Expensive
Time consuming
Risky
Difficult to satisfy the condition of homogeneity
Quantitative Forecasting Methods
Quantitative
Forecasting
Response
Cycle
Response
Summer
Response
Mo., Qtr.
Random Component
• Erratic, unsystematic, ‘residual’ fluctuations
• Due to random variation or unforeseen
events
– Union strike
– Cyclone
• Short duration &
nonrepeating
Trend Projection – Graphic Curve Fitting
Random Influences
Values of Dependent Variable
Cyclical Fluctuation
Secular Trend
Time
Trend Projection – Graphic Curve Fitting
Values of Dependent Variable
Actual
observation
Time
Trend Projection – Graphic Curve Fitting
Deviation
Values of Dependent Variable
Deviation Deviation
Actual
Deviation
Deviation observation
Deviation Point on
Deviation
the line
Time
Trend Projection – Graphic Curve Fitting
Deviation
Values of Dependent Variable
Deviation Deviation
Actual
Deviation
Deviation observation
Deviation Point on
Deviation
the line
Yˆ = a + bx
Time
Trend Projection – Graphic Curve Fitting
Assumptions
St = So + bt
• Where:
– St value of time series to be forecasted for period t
– So estimated value of time series in the base period
– b is the absolute amount of growth per period
– t time period for which series is to be forecasted
Time Series Sales Data
Time Series Sales Data
St = So + bt
Σ S = nSo + b Σ t
Σ S*t = SoΣ t + b
Σ t2
So = [(∑S )(∑t ) −(∑t )(∑S * t )] / d
2
[
b = n ∑S * t −(∑t )(∑S ) / d ]
Where: d = n ∑ ∑
t 2
− ( t ) 2
Time Series Sales Data
Period
Year Quarter (t) Sales (S) S*t t^2
1996 I 1 300 300 1
1996 II 2 305 610 4
1996 III 3 315 945 9
1996 IV 4 340 1360 16
1997 I 5 346 1730 25
1997 II 6 352 2112 36
1997 III 7 364 2548 49
1997 IV 8 390 3120 64
1998 I 9 397 3573 81
1998 II 10 404 4040 100
1998 III 11 418 4598 121
1998 IV 12 445 5340 144
n = 12 78 4376 30276 650
St = So + bt St = 281.39 + 12.81t
Σ S = nSo + b Σ t
Σ S*t = SoΣ t + b Σ t2
Sales per Quarter
500
450
400
350
300
250
200
0 2 4 6 8 10 12 14
1996 1997 1998 Quarter
Seasonal Variation
Actual
Ratio =
Trend Forecast
– Moving Averages
– Exponential Smoothing
Moving Average
Moving Average
Moving Average
Moving Average
Qua rte r Actua l M a rke t 3 Qua rte r M oving A - F (A - F)^ 2 5 Qua rte r M oving A - F (A - F)^ 2
S ha re (A) Ave ra ge (F) Ave ra ge (F)
1 20
2 22
3 23
4 24 21.67 2.33 5.44
5 18 23.00 -5.00 25.00
6 23 21.67 1.33 1.78 21.40 1.60 2.56
7 19 21.67 -2.67 7.11 22.00 -3.00 9.00
8 17 20.00 -3.00 9.00 21.40 -4.40 19.36
9 22 19.67 2.33 5.44 20.20 1.80 3.24
10 23 19.33 3.67 13.44 19.80 3.20 10.24
11 18 20.67 -2.67 7.11 20.80 -2.80 7.84
12 23 21.00 2.00 4.00 19.80 3.20 10.24
78.33 62.48
13 21.33 20.60(Forec as t)
To decide on the better moving average forecast calculate the
root-mean-square error(RMSE) of each forecast and use the
moving average which results in the smallest RMSE
RMSE = ∑ ( A −F )
t t
2
n
Moving Average
Qua rte r Actua l M a rke t 3 Qua rte r M oving A - F (A - F)^ 2 5 Qua rte r M oving A - F (A - F)^ 2
S ha re (A) Ave ra ge (F) Ave ra ge (F)
1 20
2 22
3 23
4 24 21.67 2.33 5.44
5 18 23.00 -5.00 25.00
6 23 21.67 1.33 1.78 21.40 1.60 2.56
7 19 21.67 -2.67 7.11 22.00 -3.00 9.00
8 17 20.00 -3.00 9.00 21.40 -4.40 19.36
9 22 19.67 2.33 5.44 20.20 1.80 3.24
10 23 19.33 3.67 13.44 19.80 3.20 10.24
11 18 20.67 -2.67 7.11 20.80 -2.80 7.84
12 23 21.00 2.00 4.00 19.80 3.20 10.24
78.33 62.48
13 21.33 20.60(Forec as t)
RM S E 2.95 2.99
Criticism
Ft +1 = wAt + (1 − w) Ft
0 ≤ w ≤1
Exponential Forecasting
Exponential Forecasting
Exponential Forecasting
Q ua rte r Actua l M a rke t Fore ca st w ith A-F (A - F)^ 2 Fore ca st w ith A-F (A - F)^ 2
S ha re (A) w = 0.3 w = 0.5
1 20 21.0 -1.0 1.0 21.0 -1.0 1.0
2 22 20.7 1.3 1.7 20.5 1.5 2.3
3 23 21.1 1.9 3.6 21.3 1.8 3.1
4 24 21.7 2.3 5.5 22.1 1.9 3.5
5 18 22.4 -4.4 19.0 23.1 -5.1 25.6
6 23 21.1 1.9 3.8 20.5 2.5 6.1
7 19 21.6 -2.6 7.0 21.8 -2.8 7.6
8 17 20.8 -3.8 14.8 20.4 -3.4 11.4
9 22 19.7 2.3 5.3 18.7 3.3 10.9
10 23 20.4 2.6 6.8 20.3 2.7 7.0
11 18 21.2 -3.2 10.0 21.7 -3.7 13.5
12 23 20.2 2.8 7.7 19.8 3.2 10.0
252 86.3 102.1
13 21.1 21.4
RM S E = 2.68 2.92
Evaluation of Smoothing Techniques
B
Indicator level
Coincident indicator
C
• Must be accurate
• Provide adequate lead time
• Lead time should be constant
• Logical explanation why it is a leading
indicator
• Cost and time necessary for data collection
Construction of an Index
• Indices represent a single time series made up
of a number of individual leading indicators.
– Purpose is to smooth out the random fluctuations in
each individual series.
• Composite index- weighted average of individual indicators
in each group. Good indicators are given more weightage.
– Index is interpreted in terms of percentage change from period to
period.