Business Processes
Sales Order
Management
Aggregate
Planning
Master
Scheduling
Production Activity
Control
Quality
Control
Distribution
Mngt.
© 2001 Victor E. Sower, Ph.D., C.Q.E.
Chapter 11
Forecasting
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Forecasting
• Predicting future events
• Usually demand behavior over a time frame
• Qualitative methods
– based on subjective methods
• Quantitative methods
– based on mathematical formulas
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 2
Time Frame
• Short-range to medium-range
– daily, weekly monthly forecasts of sales data
– up to 2 years into the future
• Long-range
– strategic planning of goals, products, markets
– planning beyond 2 years into the future
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 5
Demand Behavior
• Trend
– gradual, long-term up or down movement
• Cycle
– up & down movement repeating over long time frame
• Seasonal pattern
– periodic oscillation in demand which repeats
• Random movements follow no pattern
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 6
Trend
Demand
Demand
Forms Of Forecast Movement
Cycle
Random
movement
Time
Seasonal
pattern
Demand
Demand
Time
Trend with
seasonal pattern
Time
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Time
Ch 10 - 7
Forecasting Methods
• Qualitative methods
– management judgment, expertise, opinion
– use management, marketing, purchasing,
engineering
• Delphi method
– solicit forecasts from experts
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 8
Forecasting Process
1. Identify the
purpose of forecast
2. Collect
historical data
3. Plot data and
identify patterns
5. Develop / compute forecast for
period of historical data
4. Select a forecast model that
seems appropriate for data
6. Check forecast accuracy
with one or more measures
8b. Select new forecast model or
adjust parameters of existing model
7.
Is accuracy
of forecast
acceptable?
8a. Forecast over
planning horizon
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
9. Adjust forecast based
on additional qualitative
information and insight
10. Monitor results and
measure forecast accuracy
Ch 10 - 9
Time Series Methods
• Statistical methods using historical data
– moving average
– exponential smoothing
– linear trend line
• Assume patterns will repeat
• Naive forecasts
Demand
– forecast = data from last period
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 10
Forecast Accuracy
•
•
•
•
•
•
Error = Actual - Forecast
Find a method which minimizes error
Mean Absolute Deviation (MAD)
Mean Absolute Percent Deviation (MAPD)
Cumulative Error (E)
Bias
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 27
Mean Absolute Deviation (MAD)
MAD =   Dt - Ft 
n
where,
t = the period number
Dt = demand in period t
Ft = the forecast for period t
n = the total number of periods
 = the absolute value
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 28
Other Accuracy Measures
• Mean absolute percent deviation (MAPD)
MAPD 

 Dt  F t
 Dt
53.39
 0.096
520
• Cumulative errorE   et
• Average error or Bias
E
 et
n
• Mean Squared Error (MSE)
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 30
Tracking Signal
• Compute each period
• Compare to control limits
• Forecast is in control if within limits
Tracking signal 
 D t  F t 
MAD
E
MAD
MAD  0.8
Use control limits of +/- 2 to +/- 5 MAD
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 33
Monitoring Forecast Errors With
Statistical Control Charts


2
 D t  F t
n1
375.68
 6.12
10
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 36
Regression Methods
• Study relationship between two or more
variables
• Dependent variable depends on independent
variable
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 37
Linear Regression Formulas
y = a + bx
where,
a
b
x
y
=
=
=
=
b =
intercept (at period 0)
slope of the line
the independent variable
forecast for demand given x
xy - nxy
x2- nx 2
a = y-bx
where,
n = number of periods
x = x , mean of x values
n
y = y , mean of y values
n
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 38
Linear Regression Line
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 40
Correlation And Coefficient Of
Determination
• Correlation, r
– measure of strength of relationship
– varies between -1.00 and +1.00
• Coefficient of determination, r2
– percentage of variation in dependent variable
resulting form independent variable
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 41
Multiple Regression
Study relationship of demand to two or more independent
variables … where
y =  0 + 1 x 1 + 2 x 2 … + k x k
where,
0 = intercept
1, … , k = parameters for independent
variables
x1 , … , xk
= independent variables
© 2000 by Prentice-Hall Inc
Russell/Taylor Oper Mgt 3/e
Ch 10 - 43