Forecasting

1
Forecasting

• February 26, 2007

2
Laws of Forecasting

• Three Laws of Forecasting
• Forecasts are always wrong!
• Detailed forecasts are worse than aggregate
  forecasts!
• The further into the future, the less reliable
  the forecast will be!

3
Forecasting

• Starting point of all Production Planning systems
• Qualitative Forecasting techniques
• Quantitative Forecasting techniques
• Choice of technique varies with the Product Life
  Cycle

4
Product Development Stage

• Should we enter into this business? What
  segments?
• What are the alternative growth opportunities for
  product X?
• How have established products similar to X fared?
• How should we allocate RD efforts and funds?
• Where will be the market 5 years, 10 years from
  now?

5
Preliminaries

• What is the purpose of forecast? How is it to be
  used?
• Accuracy and power required by the techniques
• Requirements for entering a business vs. next
  years budget
• Impact of promotions and other marketing devices
• Techniques vary with cost, scope and accuracy
• Forecaster should fix the level of tolerance of
  accuracy
• Helps in managing the trade-offs
• Accurate forecast reduces inventory (cost of
  inventory vs. cost of forecasting)

6
Qualitative Forecasting

• Relies on expertise of people
• Data is scarce
• Usually used for technological forecasts (long
  term forecasts)
• Delphi Method, Market Research, Panel Consensus

7
Quantitative Forecasting

• Time Series models
• Predict a future parameter as a function of past
  values of that parameter (e.g., historical
  demand)
• Systematic variation is captured (seasonality,
  trend)
• Cyclic patterns
• Growth (decline) rates of the trends
• Assume future is like past (hence useful for
  short term forecasts)
• Managers need to look at the turning points in
  future that change the past trends

8
Time Series Forecasting

• Time period i 1,2,..t (most recent data)
• A(i) Actual observations
• f(t?) Forecasts for t ?, ? 1,2,,
• F(t) smoothed estimate (current position of
  the process under consideration)
• T(t) smoothed trend

Time Series Model
f(t?), ? 1,2,3,,
A(i), i 1,2,t
9
Time Series Forecasting

• Moving-Average Model
• Exponential Smoothing Model
• Exponential Smoothing with a Linear Trend Model
• Winters Method (adds seasonal multipliers to the
  exponential smoothing with linear trend model)

10
Quantitative Forecasting

• Causal models
• Most sophisticated
• Predict a future parameter (e.g., demand for a
  product) as a function of other parameters (e.g.,
  interest rates, marketing strategy).

11
Causal Forecasting

• Opening a fast food restaurant
• Demand forecast?
• Predictable parameters
• Population in the vicinity
• Competition
• Use statistics (e.g., regression) to estimate the
  parameters
• Y b0 b1x1 b2X2

12
Components of an Observation

• Observed demand (O)
• Systematic component (S) Random component (R)

Level (current deseasonalized demand)
Trend (growth or decline in demand)
Seasonality (predictable seasonal fluctuation)

• Systematic component Expected value of demand
• Random component The part of the forecast that
  deviates from the systematic component
• Forecast error difference between forecast and
  actual demand

13
Time Series Forecasting
Forecast demand for the next four quarters.
14
Time Series Forecasting
15
Basic Approach toDemand Forecasting

• Understand the objectives of forecasting
• Integrate demand planning and forecasting
• Identify major factors that influence the demand
  forecast
• Understand and identify customer segments
• Determine the appropriate forecasting technique
• Establish performance and error measures for the
  forecast

16
Patterns of Demand
(a) Horizontal Data cluster about a horizontal
line.
17
Patterns of Demand
(b) Trend Data consistently increase or decrease.
18
Patterns of Demand
(c) Seasonal Data consistently show peaks and
valleys.
19
Patterns of Demand
(c) Cyclical Data reveal gradual increases and
decreases over extended periods.
20
Demand Forecast Applications
Time Horizon Medium Term Long Term
Short Term (3 months (more than
Application (03 months) 2 years) 2 years)
Forecast quantity Individual
products or services Decision
area Inventory management Final
assembly scheduling Workforce schedulin
g Master production scheduling Forecasting T
ime series technique Causal Judgment
Total sales Groups or
families of products or services Staff
planning Production planning Mast
er production scheduling Purchasing
Distribution Causal Judgment
Total sales Facility location Capacity plann
ing Process management Causal Judgment
21
Causal MethodsLinear Regression
22
Causal MethodsLinear Regression
a 8.136 b 109.229X r 0.98 r2 0.96
23
Causal MethodsLinear Regression
24
Causal MethodsLinear Regression
25
Causal MethodsLinear Regression
26
Causal MethodsLinear Regression
27
Causal MethodsLinear Regression
28
Causal MethodsLinear Regression
29
Causal MethodsLinear Regression
30
Time-Series MethodsSimple Moving Averages
31
Time-Series MethodsSimple Moving Averages
32
Time-Series MethodsSimple Moving Averages
33
Time-Series MethodsSimple Moving Averages
34
Time-Series MethodsSimple Moving Averages
35
Time-Series MethodsSimple Moving Averages
36
Time-Series MethodsSimple Moving Averages
37
Time-Series MethodsSimple Moving Averages
38
Time-Series MethodsSimple Moving Averages
39
Time-Series MethodsSimple Moving Averages
40
Time-Series MethodsExponential Smoothing
Patient arrivals
41
Time-Series MethodsExponential Smoothing
Ft 1 Ft ? (Dt Ft )
Patient arrivals
42
Time-Series MethodsExponential Smoothing
43
Time-Series MethodsExponential Smoothing
44
Time-Series MethodsExponential Smoothing
45
Time-Series MethodsExponential Smoothing
46
Time-Series MethodsExponential Smoothing
47
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
48
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
49
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
50
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
51
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
52
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
53
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
54
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
55
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
56
Time-Series MethodsTrend-Adjusted Exponential
Smoothing
57
Time-Series MethodsSeasonal Influences
58
Time-Series MethodsSeasonal Influences
59
Time-Series MethodsSeasonal Influences
60
Seasonal Patterns
61
Seasonal Patterns
62
Choosing a MethodForecast Error
63
Choosing a MethodForecast Error
64
Choosing a MethodForecast Error
65
Choosing a MethodForecast Error
66
Choosing a MethodForecast Error
67
Choosing a MethodForecast Error
68
Choosing a MethodForecast Error
69
Choosing a MethodForecast Error
70
Choosing a MethodForecast Error
71
Choosing a MethodForecast Error
72
Summary of Learning Objectives

• What are the roles of forecasting for an
  enterprise and a supply chain?
• What are the components of a demand forecast?
• How is demand forecast given historical data
  using time series methodologies?
• How is a demand forecast analyzed to estimate
  forecast error?