Chapter 4 Forecasting

1
Chapter 4 Forecasting

• Demand behavior, approaches to forecasting,
  measures of forecast error

2
Why Forecast?

• Youre wrong more than youre right
• Often ignored or used as scapegoat
• Thankless job!
• Examples of the downside of forecasting

3
Why Forecast (the answer)

• We need to plan resources in advance!

4
Forecast accuracy

• Aggregation
• Would you rather forecast sales of all Ford
  automobiles or forecast a specific model?
• Time
• Would you rather forecast Ford sales for 2005 or
  for 2010?

5
Forecast accuracy

• Aggregation
• Rather forecast sales of all Ford automobiles or
  forecast a specific model?
• Forecasts tend to be more accurate for groups of
  items than for individual items in the group
• Time
• Rather forecast Ford sales for 2005 or for 2010?
• Forecasts tend to be more accurate for the near
  future than for the distant future

6
Common Features of Forecasts

• Forecasts often (but not always) assume that what
  happened in the past will continue in the future
• Forecasts are rarely perfect
• You are either lucky or lousy
• Forecasts tend to be more for groups of items
  than for individual items
• Forecasts tend to be more accurate for the short
  range than for the long range

7
Demand Components

• Components or Elements or Behavior
• Trend long-term linear movement up or down
• Seasonal short term recurring variations
• Cyclical long-term recurring variations
• Random Irregular doesnt fit other three
  components

8
Forecasting Approaches

• Qualitative (subjective)
• Judgment and Opinion
• Quantitative (objective)
• Associative
• External sources of data
• Historical
• Internal sources of data used

9
Judgment and Opinion - 1

• Sources
• Executives
• Marketing Sales Projections
• Customers
• Potential customers
• Experts
• Delphi method

10
Judgment and Opinion - 2

• Appropriate Use
• Irregular or random demand
• New products
• Absence of historical data
• Techniques
• Surveys, questionnaires, interviews, focus
  groups, observation
• Delphi method

11
Associative

• Sources
• External industry data
• Demographic and econometric data
• Appropriate use
• Cyclical demand
• Technique
• Leading indicator, and
• Linear regression, in conjunction with
• Correlation

12
Historical

• Sources
• Historical (time series) data
• Appropriate use
• Varies (see later slides)
• Technique types
• Multi-period pattern projection
• Single period patternless projection

13
Multi-period Pattern Projection Techniques - Trend

• Appropriate use
• Clear trend pattern over time
• Techniques
• Best fit (eyeball)
• Linear trend equation or least squares
• Yt a bt
• b n ?(ty) (?t)(?y)
• n ?t2 (?t)2
• a ?y - b ?t
• n

14
Multi-period Pattern Projection Techniques -
Seasonal

• Appropriate use
• Seasonal demand
• Related to weather, holidays, sports, school
  calendar, day of the week, etc.
• Techniques
• Seasonal indexes or relatives
• Seasonally adjusted trend
• Separate trend from seasonality

15
Single Period Patternless Projection - 1

• Appropriate use
• Lack of clear data pattern
• Limited historical data
• Techniques
• Moving Average (older method)
• Ft ?A
• n
• Weighted moving average
• Ft a(At-1) b(At-2) x(At-n)

16
Single Period Patternless Projection - 2

• Techniques (continued)
• Exponential Smoothing (newer method)
• Ft Ft-1 ?( At-1 Ft-1 )
• Naïve Forecast
• Simple (stable series)
• last periods actual (often used with
  seasonality)
• Ft At-1
• Advanced (some trend)
• Ft At-1 (At-1 - At-2)

17
Single Period Patternless Projection - 3

• Techniques (continued)
• Double exponential smoothing
• aka second order exponential smoothing
• Special case
• Incorporates some trend
• Uses exponential smoothing formula plus second
  formula with additional smoothing constant

18
Multiperiod Pattern Projection
19
Single Period Patternless Projection
20
Other Forecasting Methods
21
Measures of Forecast Error - 1

• Forecast Error (e, E, or FE)
• Et At - Ft
• Average Error (AE)
• AE ?E
• n
• Mean Absolute Deviation (MAD)
• MAD ?E
• n

22
Measures of Forecast Error - 2

• Mean Squared Error (MSE)
• MSE ?E2
• n-1
• Standard Deviation (SD)
• SD square root of ?E2
• n-1
• Mean Absolute Percent Error (MAPE)
• MAPE ?(E/A) (100)
• n

23
Controlling the forecast - 1

• Control charts
• Upper and lower control limits
• (remember SPC?) See Figure 3-11
• Formulas ___
• Upper limit 0 z vMSE
• Lower limit 0 z vMSE
• where z number of standard deviations from
  the mean

24
Controlling the forecast - 2

• Tracking Signal (TS)
• Reflects bias in the forecast
• TS ?(A F)
• MAD
• Look for values within 4

25
Choosing and

• Choosing a forecasting technique
• Nature of data (plot data pattern?)
• Forecast horizon
• Preparation time
• Experience (may want to try several)
• Choosing a measure of forecast error
• Ease of use
• Cost

26
Using

• Using forecast information
• Proactive vs. reactive
• Look at reasonability
• Assure everyone works off same data
• What if