Slides prepared by John Loucks

2002 South-Western/Thomson Learning TM

1 1

Demand Forecasting

Overview

Introduction Qualitative Forecasting Methods Quantitative Forecasting Models How to Have a Successful Forecasting System Computer Software for Forecasting Forecasting in Small Businesses and Start-Up Ventures Wrap-Up: What World-Class Producers Do

Introduction

Forecasting Estimating the future demand for products and services and the resources necessary to produce these outputs Sales forecasts Starting point for the Operations Management The sales forecasts become inputs to both business strategy and production resource forecasts.
4

Examples of Production Resource Forecasts

Forecast Horizon Long Range Medium Range Short Range Time Span Item Being Forecasted Unit of Measure Dollars, Tons Units, Pounds Units, Hours

New Products, Years Factory Capacities, Facility needs Product groups, Months Purchased materials and Inventories Days, Specific Products, Weeks Machine Capacities

Some Reasons Why Forecasting is Essential in OM

New Facility Planning It can take 5 years to design and build a new factory or design and implement a new production process. Production Planning Demand for products vary from month to month and it can take several months to change the capacities of production processes. Workforce Scheduling Demand for services (and the necessary staffing) can vary from hour to hour and employees weekly work schedules must be developed in 6 advance.

Forecasting is an Integral Part of Business Planning

Forecast Method(s) Out put: Demand Estimates

Inputs: Market conditions competitor actions, customer taste Economic Outlook - Stock Other factors

Sales Forecast

Management Team Processor Capacity, Avl.Reso, Risk, Experience

Business Strategy Marketing plan, Production plan, Finance plan

Production Resource Forecasts Long, medium and short range

7

Forecasting Methods

Qualitative Approaches Quantitative Approaches

Qualitative Approaches

Used to develop the sales forecasts

Usually based on judgments about factors that underlie the demand of particular products or services Do not require a demand history for the product or service, therefore are useful for new products/services The approach/method that is appropriate depends on a products life cycle stage
9

Qualitative Methods

Educated guess Short term forecasts Executive committee consensus Compromise forecasts Delphi method Survey of sales force - Sales forecast to ensure realistic estimates Survey of customers Historical analogy Forecasting sales for new products Market research - New products or existing products to be introduced into new market segments
10

Quantitative Forecasting Approaches

Mathematical model based on the historical data - generate the future demand based on the past data, i.e., history will tend to repeat itself Analysis of the past demand pattern provides a good basis for forecasting future demand Majority of quantitative approaches fall in the category of time series analysis Forecast accuracy High accuracy with low forecast error 11

Time Series Analysis

A time series is a set of numbers where the order or sequence of the numbers is important, e.g., historical demand

Components of a Time Series

Trends are noted by an upward or downward sloping line. Cycle is a data pattern that may cover several years before it repeats itself. Seasonality is a data pattern that repeats itself over the period of one year or less. Random fluctuation (noise) results from random variation or unexplained causes.
12

Seasonal Patterns
Length of Time Before Pattern Is Repeated Year Year Year Month Week Number of Seasons in Pattern 4 12 52 28-31 7
13

Length of Season Quarter Month Week Day Day

Quantitative Forecasting Approaches

Linear Regression and correlation Simple Moving Average Weighted Moving Average Exponential Smoothing (exponentially weighted moving average) Exponential Smoothing with Trend (double exponential smoothing)

14

Simple Linear Regression

Least square method Long term forecasting

Linear regression analysis establishes a relationship between a dependent variable and one or more independent variables in the historical observations
Ex. Independent variable time period and dependent variable - sales forecasting in sales

15

Simple Linear Regression

Regression Equation This model is of the form: Y = a + bX Y = dependent variable X = independent variable a = y-axis intercept b = slope of regression line

16

Simple Linear Regression

Constants a and b The constants a and b are computed using the following equations:
a=
2 x y- x xy

n x 2 -( x)2

b=

n xy- x y n x 2 -( x)2

17

Simple Linear Regression

Once the a and b values are computed, a future value of X can be entered into the regression equation and a corresponding value of Y (the forecast) can be calculated.

18

Example: College Enrollment

Simple Linear Regression At a small regional college enrollments have grown steadily over the past six years, as evidenced below. Use time series regression to forecast the student enrollments for the next three years. Year 1 2 3

Students Enrolled (1000s) 2.5 2.8 2.9

Year 4 5 6

Students Enrolled (1000s) 3.2 3.3 3.4

19

Example: College Enrollment

Simple Linear Regression x y x2 1 2.5 1 2 2.8 4 3 2.9 9 4 3.2 16 5 3.3 25 6 3.4 36 Sx=21 Sy=18.1 Sx2=91 xy 2.5 5.6 8.7 12.8 16.5 20.4 Sxy=66.5
20

Example: College Enrollment

Simple Linear Regression

91(18.1) 21(66.5) a 2.387 2 6(91) (21)
6(66.5) 21(18.1) b 0.180 105

Y = 2.387 + 0.180X

21

Example: College Enrollment

Simple Linear Regression Y7 = 2.387 + 0.180(7) = 3.65 or 3,650 students Y8 = 2.387 + 0.180(8) = 3.83 or 3,830 students Y9 = 2.387 + 0.180(9) = 4.01 or 4,010 students Note: Enrollment is expected to increase by 180 students per year.

22

Simple Linear Regression

Simple linear regression can also be used when the independent variable X represents a variable other than time. In this case, linear regression is representative of a class of forecasting models called causal forecasting models.

23

Example: Railroad Products Co.

Simple Linear Regression Causal Model The manager of RPC wants to project the firms sales for the next 3 years. He knows that RPCs longrange sales are tied very closely to national freight car loadings. On the next slide are 7 years of relevant historical data. Develop a simple linear regression model between RPC sales and national freight car loadings. Forecast RPC sales for the next 3 years, given that the rail industry estimates car loadings of 250, 270, and 300 million.
24

Example: Railroad Products Co.

Simple Linear Regression Causal Model Year 1 2 3 4 5 6 7 RPC Sales ($millions) 9.5 11.0 12.0 12.5 14.0 16.0 18.0 Car Loadings (millions) 120 135 130 150 170 190 220
25

Example: Railroad Products Co.

Simple Linear Regression Causal Model x 120 135 130 150 170 190 220 1,115 y 9.5 11.0 12.0 12.5 14.0 16.0 18.0 93.0 x2 14,400 18,225 16,900 22,500 28,900 36,100 48,400 185,425 xy 1,140 1,485 1,560 1,875 2,380 3,040 3,960 15,440
26

Example: Railroad Products Co.

Simple Linear Regression Causal Model

185, 425(93) 1,115(15, 440) a 0.528 2 7(185, 425) (1,115)
7(15, 440) 1,115(93) b 0.0801 2 7(185, 425) (1,115)

Y = 0.528 + 0.0801X

27

Example: Railroad Products Co.

Simple Linear Regression Causal Model Y8 = 0.528 + 0.0801(250) = $20.55 million Y9 = 0.528 + 0.0801(270) = $22.16 million Y10 = 0.528 + 0.0801(300) = $24.56 million Note: RPC sales are expected to increase by $80,100 for each additional million national freight car loadings.

28

Multiple Regression Analysis

Multiple regression analysis is used when there are two or more independent variables. An example of a multiple regression equation is: Y = 50.0 + 0.05X1 + 0.10X2 0.03X3 where: Y = firms annual sales ($millions) X1 = industry sales ($millions) X2 = regional per capita income ($thousands) X3 = regional per capita debt ($thousands)
29

Coefficient of Correlation (r)

The coefficient of correlation, r, explains the relative importance of the relationship between x and y. The sign of r shows the direction of the relationship. The absolute value of r shows the strength of the relationship. The sign of r is always the same as the sign of b. r can take on any value between 1 and +1.

30

Coefficient of Correlation (r)

Meanings of several values of r: -1 a perfect negative relationship (as x goes up, y goes down by one unit, and vice versa) +1 a perfect positive relationship (as x goes up, y goes up by one unit, and vice versa) 0 no relationship exists between x and y +0.3 a weak positive relationship -0.8 a strong negative relationship

31

Coefficient of Correlation (r)

r is computed by:
r n xy x y

2 2 2 2 n x ( x ) n y ( y )

32

Coefficient of Determination (r2)

The coefficient of determination, r2, is the square of the coefficient of correlation. The modification of r to r2 allows us to shift from subjective measures of relationship to a more specific measure. r2 is determined by the ratio of explained variation to total variation:
r2
2 ( Y y ) 2 ( y y )

33

Example: Railroad Products Co.

Coefficient of Correlation x 120 135 130 150 170 190 220 y 9.5 11.0 12.0 12.5 14.0 16.0 18.0 x2 14,400 18,225 16,900 22,500 28,900 36,100 48,400 xy 1,140 1,485 1,560 1,875 2,380 3,040 3,960 y2 90.25 121.00 144.00 156.25 196.00 256.00 324.00
34

1,115 93.0 185,425 15,440 1,287.50

Example: Railroad Products Co.

Coefficient of Correlation

7(15, 440) 1,115(93)

2 2 7(185, 425) (1,115) 7(1, 287.5) (93)

r = .9829

35

Example: Railroad Products Co.

Coefficient of Determination r2 = (.9829)2 = .966 96.6% of the variation in RPC sales is explained by national freight car loadings.

36

Ranging Forecasts

Forecasts for future periods are only estimates and are subject to error. One way to deal with uncertainty is to develop bestestimate forecasts and the ranges within which the actual data are likely to fall. The ranges of a forecast are defined by the upper and lower limits of a confidence interval.

37

Ranging Forecasts

The ranges or limits of a forecast are estimated by: Upper limit = Y + t(syx) Lower limit = Y - t(syx) where: Y = best-estimate forecast t = number of standard deviations from the mean of the distribution to provide a given probability of exceeding the limits through chance syx = standard error of the forecast
38

Ranging Forecasts

The standard error (deviation) of the forecast is computed as:

s yx =

2 y - a y - b xy

n-2

39

Example: Railroad Products Co.

Ranging Forecasts Recall that linear regression analysis provided a forecast of annual sales for RPC in year 8 equal to $20.55 million. Set the limits (ranges) of the forecast so that there is only a 5 percent probability of exceeding the limits by chance.

40

Example: Railroad Products Co.

Ranging Forecasts Step 1: Compute the standard error of the forecasts, syx.

1287.5 .528(93) .0801(15, 440) syx .5748 72 Step 2: Determine the appropriate value for t.
n = 7, so degrees of freedom = n 2 = 5. Area in upper tail = .05/2=0.025 Area in lower tail = .05/2=0.025 Appendix B, Table 2 shows t = 2.571.

41

Example: Railroad Products Co.

Ranging Forecasts Step 3: Compute upper and lower limits.

Upper limit = 20.55 + 2.571(.5748) = 20.55 + 1.478 = 22.028 Lower limit = 20.55 - 2.571(.5748) = 20.55 - 1.478 = 19.072 We are 95% confident the actual sales for year 8 will be between $19.072 and $22.028 million.
42

Short-Range Forecasts

Time spans ranging from a few days to a several weeks Cycles, seasonality, and trend may have little effect Random fluctuation is main data component
Ex: How much inventory of a particular product should be carried next month? How much raw materials is required to make a product to deliver on next week?

43

Evaluating Forecast-Model Performance

Short-range forecasting models are evaluated on the basis of three characteristics: Impulse response Noise-dampening ability Accuracy

44

Evaluating Forecast-Model Performance

Impulse Response and Noise-Dampening Ability If forecasts have little period-to-period fluctuation, they are said to be noise dampening. Forecasts that respond quickly to changes in historical data are said to have a high impulse response. Similarly, it has little impact on the historical data than its low impulse response. A forecast system that responds quickly to data changes necessarily picks up a great deal of random fluctuation (noise). Hence, there is a trade-off between high impulse 45 response and high noise dampening.

Evaluating Forecast-Model Performance

Accuracy Accuracy is the typical criterion for judging the performance of a forecasting approach Accuracy is how well the forecasted values match the actual values

Accuracy can be measured in several ways Standard error of the forecast (Syx) (discussed earlier) Mean absolute deviation (MAD) Mean squared error (MSE)
46

Monitoring Accuracy

Mean Absolute Deviation (MAD)

Sum of absolute deviation for n periods MAD = n

MAD =

Actual demand -Forecast demand

i i=1

47

Monitoring Accuracy

Mean Squared Error (MSE) MSE = (Syx)2 A small value for Syx means data points are tightly grouped around the line and error range is small. When the forecast errors are normally distributed, the values of MAD and syx are related:

MSE = 1.25 (MAD)

48

Short-Range Forecasting Methods

(Simple) Moving Average Weighted Moving Average Exponential Smoothing Exponential Smoothing with Trend

49

Simple Moving Average

Average the data from a few recent periods, and this average becomes the forecast for the next period An averaging period (AP) is given or selected It is called a simple average because each period used to compute the average is equally weighted It is called moving because as new demand data becomes available, the oldest data is not used . . . more
50

Weighted Moving Average

This is a variation on the simple moving average where the weights used to compute the average are not equal. This allows more recent demand data to have a greater effect on the moving average The weights must add to 1.0 and generally decrease in value with the age of the data. The distribution of the weights determine the impulse response of the forecast. . . . more
51

Example: Short term Forecasting

Moving Average Representative Historical Data

Week 1 2 3 4 5 6 7 8 9

Act.Inven.Demand 100 10 125 11 90 12 110 13 105 14 130 15 85 16 102 17 110

90 105 95 115 120 80 95 100

52

Moving Average Short-Range Forecasting

Week 8 AP = 3 weeks 106.7 Forecasts 5 weeks 104.0 7 weeks 106.4

9
10 11 12 13

105.7
99 100.7 101.7 96.7

106.4
106.4 103.4 98.4 100.4

106.7
104.6 104.6 103.9 102.4

14
15 16 17

105
110 105 98.3

103
105 103 101

100.3
105.3 102.1 100 53

Moving Average

Sample computations Considering a 10th week F3 = 85+102+110/3=99 F5 = 85+102+110+130+85/5=106.4 F7 = 90+110+105+130+85+102+110/7=104.6

54

Mean Absolute Deviation

Week Actual 8 9 10 11 12 13 14 15 16 17 Demand AP = 3 weeks 102 106.7 4.7 110 105.7 4.3 90 99 9 105 100.7 4.3 95 101.7 6.7 115 96.7 18.3 120 105 15 80 110 30 95 105 10 100 98.3 1.7 Forecasts 5 weeks 104.0 106.4 106.4 103.4 98.4 100.4 103 105 103 101 7 weeks 106.4 106.7 104.6 104.6 103.9 102.4 100.3 105.3 102.1 100

MAD

104/10 = 10.4

9.26

9.63
55

Moving Average Forecast vs Actual Cash Demand

56 Forecast for 18th week = 115+120+80+95+100/5 = 102

Exponential Smoothing

The weights used to compute the forecast (moving average) are exponentially distributed. The forecast is the sum of the old forecast and a portion (a) of the forecast error (A t-1 - Ft-1). Ft = Ft-1 + a (A t-1 - Ft-1)

. . . More Smoothing constant (a), must be between 0.0 and 1.0. A large a provides a high impulse response forecast. A small a provides a low impulse response forecast.
57

Exponential Smoothing Short-Range Forecast

Week 7 8 9 10 11 12 13 14 15 16 17 Actual Demand 85 102 110 90 105 95 115 120 80 95 100 Forecasts = 0.1 85 85 86.7 89 = 0.2 85 85 88.4 92.7 = 0.3 85 85 90.1 96.1

94.6

97.7

98

Sample computations Considering a 10th week forecasts - Ft = Ft-1 + a (A t-1 - Ft-1) = 0.1; F10 = 86.7+0.1(110-86.7) = 89 = 0.2; F10 = 88.4+0.2(110-88.4) = 92.7 = 0.3; F10 = 90.1+0.3(110-90.1) = 96.1

58

Mean Absolute Deviation

Week 8 9 10 11 12 13 14 15 16 17 Actual Demand 102 110 90 105 95 115 120 80 95 100 Forecasts = 0.1 = 0.2 85 (17) 85 (17) 86.7 (23.3) 88.4 (21.6) 89 (1) 92.7 (2.7) = 0.3 85 (17) 90.1 (19.9) 96.1 (6.1)

94.6 (5.4)

97.7 (2.3)

98 (2.0)

MAD

133.9/10 = 13.9

12.44

12.60

59

Exponential Smoothing Forecast vs Actual Cash Demand

60

Forecast for 18th week (= 0.2 ) = Ft = Ft-1 + (A t-1 - Ft-1) = F17 + (A 17 F17) F18 = 97.7 + 0.2 (100- 97.7) = 98.2

61

End of Chapter 3

62