Forecasting

1
Forecasting

• Ross L. Fink

2
Forecasting

• Definition--Forecasting tries to predict the
  future
• Used for planning purposes
• Forecasting goes beyond predicting demand
• Internal manufacturing costs
• Cost of raw materials and component parts
• Cost and availability of energy
• Interest rates
• Stock prices

3
Rules of Forecasting

• Your forecast is never correct
• Forecasts assume some underlying causal system
• Generally, the shorter the time horizon of the
  forecast, the more accurate the forecast
• Generally, forecasting aggregations (groups of
  items) is more accurate than forecasting
  individual items

4
Forecasting Methods

• Qualitative
• Quantitative

5
Qualitative Forecasting Methods

• Consumer surveys
• Test marketing
• Sales force composite
• Executive opinion
• Delphi technique
• Panel of experts

6
Quantitative Forecasting Methods

• Time series
• Naïve
• Moving averages
• Exponential smoothing
• Box-Jenkins
• Associative or causal models
• Simple regression
• Multiple regression

7
Decomposition of Time Series Data

• Believe that data consists of components
• Trend
• Seasons
• Cycles
• Random and irregular variations
• Approach--isolate components and forecast
  separately and then recombine to obtain final
  forecast

8
Ways to Select Forecasting Method

• Noise-dampening (smoothing)
• Response (lag)
• Error measurement

9
Error Measurement

• Mean Absolute Deviation (MAD)--Most popular
• Average Error (AE) or bias
• Mean Square Error (MSE)
• Mean Absolute Percent Error (MAPE)
• Standard deviation

10
Error

• where,
• Et error for period t
• At actual demand for period t
• Ft forecast for period t

11
Mean Absolute Deviation
12
Example
13
Example
14
Example--3-month Moving Average
15
Example
16
Example
17
Simple Exponential Smoothing

• OR
• Alpha (smoothing Constant) is between 0 and 1