Production and Operations Management

Chapter 8
Forecasting
Learning Goals (1 of 2)

8.1 Explain how managers can change demand patterns.

8.2 Describe the two key decisions on making forecasts.
8.3 Calculate the five basic measures of forecast errors.
8.4 Compare and contrast the four approaches to
judgmental forecasting.
Learning Goals (2 of 2)

8.5 Use regression to make forecasts with one or more

independent variables.
8.6 Make forecasts using the five most common statistical
approaches for time-series analysis.
What is a Forecast?

• Forecast
– A prediction of future events used for planning
purposes.
Demand Patterns (1 of 5)

• Time series
– The repeated observations of demand for a service or
product in their order of occurrence
• There are five basic time series patterns
1. Horizontal: The fluctuation of data around a constant mean
2. Trend: The systematic increases or decreases in the mean of
the series over time
3. Seasonal: A repeatable pattern or increases or decreases in
demand, depending on the time of day, week, month, or season
4. Cyclical: The less predictable gradual increases in demand over
longer periods of time
5. Random: The unforecastable variation in demand
Demand Patterns (2 of 5)
Figure 8.1 Patterns of Demand
(a) Horizontal: Data cluster about a horizontal line
Demand Patterns (3 of 5)
Figure 8.1 [continued]
(b) Trend: Data consistently increase or decrease
Demand Patterns (4 of 5)
Figure 8.1 [continued]
(c) Seasonal: Data consistently show peaks and valleys
Demand Patterns (5 of 5)
Figure 8.1 [continued]
(d) Cyclical: Data reveal gradual increases and decreases over
extended periods
Key Decisions on Making Forecasts

• Deciding What to Forecast

– Level of aggregation
– Units of measurement
• Choosing the Type of Forecasting Technique
– Judgment methods
– Causal methods
– Time-series analysis
– Trend projection using regression
Forecast Error

• For any forecasting method, it is important to measure the

accuracy of its forecasts.
• Forecast error is simply the difference found by subtracting
the forecast from actual demand for a given period, or
Et = Dt − Ft
where
Et = forecast error for period t
Dt = actual demand in period t
Ft = forecast for period t
Measures of Forecast Error (1 of 3)

Cumulative sum of Standard deviation

forecast errors (Bias)
 
2
 Et  E
CFE   Et  
n 1
Average forecast error Mean Absolute Deviation

E
CFE  Et
MAD 
n n

Mean Squared Error Mean Absolute Percent Error

 Et
2
MAPE 
 E t Dt  100 
MSE 
n n
Measures of Forecast Error (2 of 3)
Figure 8.2(B) Detailed Calculations of Forecast Errors
Blank Actual Forecast Error Absolute
Error Error Error
Error ^2cap 2 Pct Error Pct
Absolute
Error

Past period 1 39 41 −2 2 4 5.128%

Past period 2 37 43 −6 6 36 16.216%
Past period 3 55 45 10 10 100 18.182%
Past period 4 40 50 −10 10 100 25%
Past period 5 59 51 8 8 64 13.559%
Past period 6 63 56 7 7 49 11.111%
Past period 7 41 61 −20 20 400 48.78%
Past period 8 57 60 −3 3 9 5.236%
Past period 9 56 62 −6 6 36 10.714%
Past period 10 54 63 −9 9 81 16.667%
Totals 501 Blank −31 81 879 170.621%
Average 50.1 Blank −3.1 8.1 87.9 17.062%
Next period forecast Blank 0 (Bias) (MAD) (MSE) (MAPE)
Blank Blank Blank Blank std err 9.883 Blank
Measures of Forecast Error (3 of 3)
Figure 8.2(C) Error Measures

Measure Value
Error Measures Blank
CFC (Cumulative Forecast Error) −31
MAD (Mean Absolute Deviation) 8.1
MSE (Mean Squared Error) 87.9
Standard Deviation of Errors 9.883
MAPE (Mean Absolute Percent Error) 17.062%
Example 1 (1 of 4)
The following table shows the actual sales of upholstered chairs for a
furniture manufacturer and the forecasts made for each of the last eight
months.
Calculate CFE, MSE, σ, MAD, and MAPE for this product.
Example 1 (2 of 4)
Using the formulas for the measures, we get:
Cumulative forecast error (mean bias)
CFE = −15
Average forecast error (mean bias):
CFE 15
E   1.875
n 8
Mean squared error:

 Et
2
5,275
MSE    659.4
n 8
Example 1 (3 of 4)
Standard deviation:

 Et  ( 1.875)
2

 =  27.4
n 1

Mean absolute deviation:

 Et 195
MAD = = = 24.4
n 8

Mean absolute percent error:

MAPE =
 E t Dt  100 
=
81.3%
= 10.2%
n 8
Example 1 (4 of 4)
• A CFE of −15 indicates that the forecast has a slight bias to
overestimate demand.
• The MSE, σ, and MAD statistics provide measures of forecast error
variability.
• A MAD of 24.4 means that the average forecast error was 24.4 units
in absolute value.
• The value of σ, 27.4, indicates that the sample distribution of forecast
errors has a standard deviation of 27.4 units.
• A MAPE of 10.2 percent implies that, on average, the forecast error
was about 10 percent of actual demand.
These measures become more reliable as the number of periods of
data increases.
Judgment Methods

• Other methods (casual, time-series, and trend projection

using regression) require an adequate history file, which
might not be available.
Causal Methods: Linear Regression
• Dependent variable – The variable that one wants to forecast
• Independent variable – The variable that is assumed to affect
the dependent variable and thereby “cause” the results
observed in the past
• Simple linear regression model is a straight line
Y = a + bX
where
Y = dependent variable
X = independent variable
a = Y-intercept of the line
b = slope of the line
Linear Regression (1 of 2)
Figure 8.3 Linear Regression Line Relative to Actual Demand
Linear Regression (2 of 2)
• The sample correlation coefficient, r
– Measures the direction and strength of the relationship between
the independent variable and the dependent variable.
– The value of r can range from 1.00  r  1.00
• The sample coefficient of determination, r 2
– Measures the amount of variation in the dependent variable
about its mean that is explained by the regression line
– The values of r 2 rangefrom0.00  r ²  1.00
• The standard error of the estimate, syx
– Measures how closely the data on the dependent variable
cluster around the regression line
Example 2 (1 of 4)
The supply chain manager seeks a better way to forecast the demand
for door hinges and believes that the demand is related to advertising
expenditures. The following are sales and advertising data for the past
5 months:
Month Sales (thousands of units) Advertising (thousands of $)
1 264 2.5
2 116 1.3
3 165 1.4
4 101 1.0
5 209 2.0

The company will spend $1,750 next month on advertising for the
product. Use linear regression to develop an equation and a forecast
for this product.
Example 2 (2 of 4)

We used POM for Windows to determine the best values

of a, b, the correlation coefficient, the coefficient of
determination, and the standard error of the estimate
a = −8.135
b = 109.229
r = 0.980
r 2  0.960
syx = 15.603
The regression equation is Y = −8.135 + 109.229X
Example 2 (3 of 4)
The r of 0.98 suggests an unusually strong positive relationship
between sales and advertising expenditures. The coefficient of
determination, r 2 , implies that 96 percent of the variation in sales
is explained by advertising expenditures.

Figure 8.4 Linear

Regression Line
for the Sales and
Advertising Data
Using POM for
Windows
Example 2 (4 of 4)

• Forecast for month 6:

Y = −8.135 + 109.229X
Y  8.135  109.229 1.75 
Y = 183.016 or 183,016 units
Time Series Methods

• Naïve forecast
– The forecast for the next period equals the demand
for the current period (Forecast = Dt)
• Horizontal Patterns: Estimating the average
– Simple moving average
– Weighted moving average
– Exponential smoothing
Simple Moving Averages

• Specifically, the forecast for period t + 1 can be

calculated at the end of period t (after the actual demand
for period t is known) as

Sum of last n demands Dt + Dt 1 + Dt 2 +    + Dt n+1

Ft 1  
n n

where
Dt = actual demand in period t
n = total number of periods in the average
Ft+1 = forecast for period t + 1
Example 3 (1 of 2)
a. Compute a three-week moving average forecast for the
arrival of medical clinic patients in week 4. The numbers
of arrivals for the past three weeks were as follows:
Week Patient Arrivals
1 400
2 380
3 411

b. If the actual number of patient arrivals in week 4 is 415,

what is the forecast error for week 4?
c. What is the forecast for week 5?
Example 3 (2 of 2)

a. The moving average forecast at Week Patient Arrivals

1 400
the end of week 3 is:
2 380
411  380  400 3 411
F4   397.0
3
b. The forecast error for week 4 is
E4  D4  F4  415  397  18

c. The forecast for week 5 requires the actual arrivals from

weeks 2 through 4, the three most recent weeks of data
415  411  380
F5   402.0
3
Weighted Moving Averages

In the weighted moving average method, each historical

demand in the average can have its own weight, provided
that the sum of the weights equals 1.0.
The average is obtained by multiplying the weight of each
period by the actual demand for that period, and then
adding the products together.

Ft 1  W1D1  W2D2      Wn Dt n 1
where

W1 > W2 > … > Wn

Example 4 (1 of 2)
Using the following customer-arrival data, let W1 = 0.50, W2 = 0.30, and
W3 = 0.20.

a) Use the weighted moving average method to forecast arrivals for

month 5.
b) Given the number of customers that actually arrived (805), what is
the forecast error?
c) If the actual number of arrivals in month 5 is 805, compute the
forecast for month 6:
Example 4 (2 of 2)
a) The weighted moving average forecast for week 5 is:

F5 = W1D4 + W2D3 + W3D2

= 0.50(790) + 0.30(810) + 0.20(740) = 786

b) The forecast error for week 5 is:

E5 = D5 – F5 = 805 – 786 = 19

c) The weighted moving average forecast for week 6 is:

F6 = W 1 D 5 + W 2 D 4 + W 3 D 3
= 0.50(805) + 0.30(790) + 0.20(810) = 801.5
Exponential Smoothing (1 of 2)

• A sophisticated weighted moving average that calculates

the average of a time series by implicitly giving recent
demands more weight than earlier demands
• Requires only three items of data
– The last period’s forecast
– The actual demand for this period
– A smoothing parameter, alpha (α), where 0  α  1.0
• The equation for the forecast is
Ft 1  α Demand this period   (1   )(Forecast calculated last period)
 αDt  (1   )Ft
Exponential Smoothing (2 of 2)

• The emphasis given to the most recent demand levels

can be adjusted by changing the smoothing parameter.
• Larger α values emphasize recent levels of demand and
result in forecasts more responsive to changes in the
underlying average.
• Smaller α values are analogous to increasing the value of
n in the moving average method and giving greater
weight to past demand.
Example 5 (1 of 3)

a. Reconsider the patient arrival data in Example 3. It is

now the end of week 3 so the actual arrivals is known to
be 411 patients. Using α = 0.10, calculate the
exponential smoothing forecast for week 4.
b. What was the forecast error for week 4 if the actual
demand turned out to be 415?
c. What is the forecast for week 5?
Example 5 (2 of 3)

a. To obtain the forecast for week 4, using exponential

smoothing with and the initial forecast of 390*, we
calculate the forecast for week 4 as:
F 4  0.10  411  0.90  390   392.1

Thus, the forecast for week 4 would be 392 patients.

Example 5 (3 of 3)

b. The forecast error for week 4 is

E4 = 415 − 392 = 23
c. The new forecast for week 5 would be
F5  0.10  415   0.90  392.1  394.4 or 394 patients.
Trend Patterns: Using Regression

• A trend in a time series is a systematic increase or

decrease in the average of the series over time
• The forecast can be improved by calculating an estimate
of the trend
• Trend Projection with Regression accounts for the trend
with simple regression analysis.
• The regression equation is:
Ft = a + bt
Example 6 (1 of 4)

• Medanalysis, Inc., provides medical laboratory services.

Managers are interested in forecasting the number of
blood analysis requests per week. There has been a
national increase in requests for standard blood tests. The
arrivals over the next 16 weeks are given in Table 8.1.
• What is the forecasted demand for the next three periods?
Example 6 (2 of 4)
Table 8.1 Arrivals At Medanalysis for Last 16 Weeks

Week Arrivals Week Arrivals

1 28 9 61
2 27 10 39
3 44 11 55
4 37 12 54
5 35 13 52
6 53 14 60
7 38 15 60
8 57 16 75
Example 6 (3 of 4)
Figure 8.6(a) Trend Projection with Regression Results
Example 7 (4 of 4)
Figure 8.6(b) Detailed Calculations of Forecast Errors
Seasonal Patterns: Using Seasonal Factors

• Multiplicative seasonal method

– A method whereby seasonal factors are multiplied by
an estimate of average demand to arrive at a
seasonal forecast.
• Additive seasonal method
– A method in which seasonal forecasts are generated
by adding a constant to the estimate of average
demand per season.
Multiplicative Seasonal Method

Multiplicative seasonal method:

1. For each year, calculate the average demand for each
season by dividing annual demand by the number of
seasons per year.
2. For each year, divide the actual demand for each
season by the average demand per season, resulting in
a seasonal factor for each season.
3. Calculate the average seasonal factor for each season
using the results from Step 2.
4. Calculate each season’s forecast for next year.
Example 7 (1 of 3)

The manager of the Stanley Steemer carpet cleaning

company needs a quarterly forecast of the number of
customers expected next year. The carpet cleaning business
is seasonal, with a peak in the third quarter and a trough in
the first quarter.
The manager wants to forecast customer demand for each
quarter of year 5, based on an estimate of total year 5
demand of 2,600 customers.
The table on the following slides shows the quarterly
demand data from the past 4 years along with the calculation
of the seasonal factor for each week.
Example 7 (2 of 3)

Average
Seasonal
Factor

0.2043

1.2979

2.0001

0.4977
Example 7 (3 of 3)

Quarterly Forecasts

Quarter Forecast
1 650 
650 times 0.2043=132.795
0.2043 132.795

2 650times
650 1.29791.2979= 843.635
843.635

3 650times
650 2.001 = 11,300.065
2.001 ,300.065
4 650times
650 0.4977= 323.505
0.4977 323.505
Criteria for Selecting Time-Series Method
• Criteria
1. Minimizing bias (CFE)
2. Minimizing MAPE, MAD, or MSE
3. Maximizing r 2 for trend projections using regression
4. Using a holdout sample analysis
5. Using a tracking signal
6. Meeting managerial expectations of changes in the
components of demand
7. Minimizing the forecast errors in recent periods
Choosing a Time-Series Method (1 of 2)

• Using Statistical Criteria

1. For projections of more stable demand patterns, use
lower α values or larger n values to emphasize
historical experience.
2. For projections of more dynamic demand patters, use
higher α values or smaller n values.
Choosing a Time-Series Method (2 of 2)

• Holdout sample
– Actual demands from the more recent time periods in
the time series that are set aside to test different
models developed from the earlier time periods.
Tracking Signals (1 of 3)
• A tracking signal is a measure that indicates whether a
method of forecasting is accurately predicting actual
changes in demand.

CFE CFE
Tracking signal = or
MAD MADt

Each period, the CFE and MAD are updated to reflect

current error, and the tracking signal is compared to some
predetermined limits.
Tracking Signals (2 of 3)

• The MAD can be calculated as the simple average of all

absolute errors or as a weighted average determined by
the exponential smoothing method
MADt   | Et |  (1  )MADt 1

If forecast errors are normally distributed with a mean of 0,

the relationship between σ and MAD is simple
  
    MAD   1.25 MAD 
 2
MAD  0.7978  0.8 where   3.1416
Using Multiple Forecasting Methods

• Combination forecasts
– Forecasts that are produced by averaging
independent forecasts based on different methods,
different sources, or different data
• Focus forecasting
– A method of forecasting that selects the best forecast
from a group of forecasts generated by individual
techniques.
Solved Problem 1 (1 of 2)
Chicken Palace periodically offers carryout five-piece chicken dinners
at special prices. Let Y be the number of dinners sold and X be the
price. Based on the historical observations and calculations in the
following table, determine the regression equation, correlation
coefficient, and coefficient of determination. How many dinners can
Chicken Palace expect to sell at $3.00 each?
Observation Price (X) Dinners Sold (Y)
1 $2.70 760
2 $3.50 510
3 $2.00 980
4 $4.20 250
5 $3.10 320
6 $4.05 480
Total $19.55 3,300
Average $3.26 550
Solved Problem 1 (2 of 2)
We use the computer to calculate the best values of a, b, the
correlation coefficient, and the coefficient of determination
a = 1,454.60
b = −277.63
r = −0 .84
r ²  0.71
The regression line is
Y = a + bX = 1,454.60 − 277.63X
For an estimated sales price of $3.00 per dinner
Y  a  bX  1,454.60  277.63  3.00 
 621.71 or 622dinners
Solved Problem 2 (1 of 3)
The Polish General’s Pizza Parlor is a small restaurant catering to
patrons with a taste for European pizza. One of its specialties is Polish
Prize pizza. The manager must forecast weekly demand for these special
pizzas so that he can order pizza shells weekly. Recently, demand has
been as follows:
Week Pizzas Week Pizzas
June 2 50 June 23 56
June 9 65 June 30 55
June 16 52 July 7 60

a. Forecast the demand for pizza for June 23 to July 14 by using the
simple moving average method with n = 3 then using the weighted
moving average method with weights of 0.50, 0.30, and 0.20, with .50
applying to the most recent demand.
b. Calculate the MAD for each method.
Solved Problem 2 (2 of 3)

a. The simple moving average method and the weighted

moving average method give the following results:
Current Simple Moving Average Weighted Moving Average Forecast for Next Week
Week Forecast for Next Week
52 plus 65 plus 50, over 3 left bracket left paranthesis 0.5 times 52 right paranthesis + left
June 16 52 + 65
equals + 50
55.7 or 56.
= 55.7 or 56 
 0.5  52
parantesis  
0.3 times  
 0.365 right 
0.2  50 + left
65 paranthesis  55.5
paranthesis
or 56 0.2
3 times 50 right paranthesis right bracket = 55.5 or 56
56 plus 52 plus 65, over 3 left bracket left paranthesis 0.5 times 56 right paranthesis + left
June 23 56 + 52
equals + 65
57.7 or 58.
= 57.7 or 58 
 0.5  56
parantesis  
0.3 times  
 0.352 right 
0.2  65 + left
52 paranthesis  56.6 or 57 0.2
paranthesis
3 times 65 right paranthesis right bracket = 56.6 or 57
55
55plus
+ 5656+ plus
52 52, over 3
June 30 equals 54.3 or = 54.3 or 54
54. 
parantesis     
left bracket left paranthesis 0.5 times 55 right paranthesis + left
 0.5  55  0.3  56  0.2  52   54.7 or 55
0.3 times 56 right paranthesis +left paranthesis 0.2
3 times 52 right paranthesis right bracket = 54.7 or 55
60
60plus
+ 5555+ plus
56 56, over 3
July 7 = 57.0 or 57
equals 57.0 or 57.
     
left bracket left paranthesis 0.5 times 60 right paranthesis + left
 0.5  60  0.3  55  0.2  56   57.7 or 58
parantesis 0.3 times 55 right paranthesis + left paranthesis 0.2
3 times 56 right paranthesis right bracket = 57.7 or 58
Solved Problem 2 (3 of 3)
b. The mean absolute deviation is calculated as follows:

For this limited set of data, the weighted moving average method resulted
in a slightly lower mean absolute deviation. However, final conclusions
can be made only after analyzing much more data.
Solved Problem 3 (1 of 4)
The monthly demand for units manufactured by the Company has been as
follows:
Month Units Month Units
May 100 September 105
June 80 October 110
July 110 November 125
August 115 December 120

a. Use the exponential smoothing method to forecast June to January. The

initial forecast for May was 105 units; α = 0.2.

b. Calculate the absolute percentage error for each month from June through
December and the MAD and MAPE of forecast error as of the end of
December.

c. Calculate the tracking signal as of the end of December. What can you say
about the performance of your forecasting method?
Solved Problem 3 (2 of 4)
a.
Current Month, Calculating Forecast for Next Month Forecast for
t Ft 1   Dt  (1   )Ft Month t + 1

May 0.20.2 100

times  + 0.8
100 105105
0.8times   =104.0
104.0or
or 104
104 June

June 0.20.2  8080

times  + 0.8 104.0
0.8times   =99.2
104.0 99.2 or 99
or 99 July

July 0.20.2 110

times  + 0.8
110 99.299.2
0.8times   =101.4
101.4 or
or 101
101 August

August 0.20.2 115115

times  + 0.8 101.4
0.8times  = 104.1
101.4 or 104
104.1 or 104 September

September 0.20.2 105105

times  + 0.8 104.1
0.8times   =104.3
104.1 or 104
104.3 or 104 October

October 0.20.2 110110

times  + 0.8 104.3
0.8times  =105.4
104.3 or105
105.4 or 105 November

November 0.20.2 125125

times  + 0.8 105.4
0.8times   109.3
105.4 = 109.3or
or109
109 December

December 0.2
0.2 120120
times  + 0.8 109.3109.3
0.8times   111.4
= 111.4or
or111
111 January
Solved Problem 3 (3 of 4)
b.

MAD =
 Et
=
87
= 12.4 MAPE =
 E t Dt  100 
=
83.7%
= 11.96%
n 7 n 7
Solved Problem 3 (4 of 4)

c. As of the end of December, the cumulative sum of

forecast errors (CFE) is 39. Using the mean absolute
deviation calculated in part (b), we calculate the tracking
signal:
CFE 39
Tracking signal = = = 3.14
MAD 12.4

The probability that a tracking signal value of 3.14 could be

generated completely by chance is small. Consequently,
we should revise our approach. The long string of forecasts
lower than actual demand suggests use of a trend method.
Solved Problem 4 (1 of 3)
The Northville Post Office experiences a seasonal pattern of daily mail volume
every week. The following data for two representative weeks are expressed in
thousands of pieces of mail:

Day Week 1 Week 2

Sunday 5 8
Monday 20 15
Tuesday 30 32
Wednesday 35 30
Thursday 49 45
Friday 70 70
Saturday 15 10
Total 224 210

a. Calculate a seasonal factor for each day of the week.

b. If the postmaster estimates 230,000 pieces of mail to be sorted next week,

forecast the volume for each day.
Solved Problem 4 (2 of 3)
Solved Problem 4 (3 of 3)

b. The average daily mail volume (in pieces of mail) is expected

230,000
to be = 32,857
7
Using the average seasonal factors calculated in part (a),
we obtain the following forecasts: