Chapter Two: Forecasting
Chapter Two: Forecasting
Chapter Two: Forecasting
Forecasting
Planning
Forecast
Customer
Production Finished
Inputs Process Goods
Forecasting
Marketing: forecasts sales for new and
existing products.
Exponential
Smoothing
Seasonality
Models
Types of forecasting Methods
Subjective methods
Sales force composites
Customer survey
Jury of executive opinion
The Delphi method
Objective methods
Causal methods
Time series methods
Qualitative Methods
Dont have data
Dont have time to develop a forecast
Often used in practice
Close enough
Depend on expert opinions
Market surveys
More appropriate for long term forecasts
Delphi Technique
A method to obtain a consensus forecast
by using opinions from a group of
experts
expert opinion
consulting salespersons
consulting consumers
Causal Methods
Causal methods use data from sources other than the
series being predicted.
Y = f (X1 , X2 , . .., Xn )
1200
1000
800
600
400
200
Period
Building Models
Plot data over time. (remove outliers & get right scale).
Using part of the data, estimate model parameters.
Forecast the rest of the data with the model.
Evaluate accuracy of the model.
Use judgment to modify.
Keep track of model accuracy over time (redo, if needed).
Forecasting Stationary
Series
Time series Analysis
Patterns that arise most often
Trend
Seasonality
Cycles
Randomness
Fig. 2-2
Time Series Patterns
Notation
{Dt , t 1} :
forecast made for period t in period t-1
forecast made at the end of t-1 after having
Ft : observed , ,
Dt 1 Dt 2
Time Series Forecast
Ft an Dt n
n 0
a0 , a1 ,....
Evaluating forecasts
Forecast error in period t
et Ft Dt
For multiple-step-ahead
et Ft Dt
Evaluating Forecasts
Mean Absolute Deviation
n
| e | i
MAD i 1
n
Mean Square Error
n
e 2
i
MSE i 1
n
Fig. 2-3
D i
Dt 1 Dt 2 ... Dt N
Ft i t N
N N
Exponential Smoothing
Ft Dt 1 (1 ) Ft 1
Moving Average
Dt 1 Dt 2 ... Dt N
Ft
N
Month Deliveries
Jan 120
Feb 90
Mar 100
Apr 75
May 110
Jun 50
Jul 75
Aug 130
Sep 110
Oct 90
Month Deliveries MA(3) MA(6)
Jan 120
Feb 90
Mar 100
Apr 75 103
May 110 88
Jun 50 95
Jul 75 78 91
Aug 130 78 83
Sep 110 85 90
Oct 90 105 92
110 94
Month Deliveries MA(3) MA(6)
1 2
2 4
3 6
4 8 4
5 10 6
6 12 8
7 14 10 7
8 16 12 9
9 18 14 11
10 20 16 13
11 22 18 15
12 24 20 17
22 19
Fig. 2-4
Moving-Average Forecasts
Lag Behind a Trend
EXPONENTIAL SMOOTHING
Current forecast is a weighted average of the
last forecast and the current value of demand
Ft Dt 1 (1 ) Dt 2 (1 ) 2 Ft 2
Ft 1 Dt i 1
i
i o
Example
Weights in Exponential
Smoothing
Fig. 2-6
1 N 1
2
Similarities & Differences
Stationary series ES weighted average
Single parameter of all past data
Lag behind a trend MA only last N periods
When =2/(N+1)
Ft a bt
Double Exponential
Smoothing
Ft ,t St Gt
Double Exponential Smoothing
Intercept at time t
St Dt (1 )( St 1 Gt 1 )
and slope at time t
Gt ( St St 1 ) (1 )Gt 1
Ft ,t St Gt