Operations Management

Forecasting

1
What is Forecasting?

2
 Forecasting at Disney World

• Global portfolio includes parks in Shanghai, Hong Kong, Paris,

Tokyo, Orlando, and Anaheim
• Revenues are derived from people – how many visitors and
how they spend their money
• Daily management report contains only the forecast and actual
attendance at each park

ENMG 605: Operations Management 3

Disney’s global portfolio includes Hong Kong Disneyland (opened 2005), Disneyland Paris (1992), and Tokyo Disneyland (1983). But it is Walt Disney World Resort
(in Florida) and Disneyland Resort (in California) that drive profits in this $40 billion corporation.
Revenues at Disney are all about people—how many visit the parks and how they spend money while there.
The daily management report contains only the forecast of yesterday’s attendance at the parks and the actual attendance. An error close to zero is
expected.

3
 Forecasting at Disney World

• Disney generates daily, weekly, monthly, annual, and 5-year

forecasts
• Forecast used by labor management, maintenance,
operations, finance, and park scheduling
• Forecast used to adjust opening times, rides, shows, staffing
levels, and guests admitted

ENMG 605: Operations Management 4

The forecasting team at Walt Disney World Resort doesn’t just do a daily prediction, however, and Iger is not its only customer. The team also provides daily,
weekly, monthly, annual, and 5-year forecasts to the labor management, maintenance, operations, finance, and park scheduling departments. Forecasters use
judgmental models, econometric models, moving-average models, and regression analysis.

Attendance forecasts for the parks drive a whole slew of management decisions. The Forecast is used to adjust opening times, rides, shows,
staffing levels, and guests admitted

4
 Forecasting at Disney World

• 20% of customers come from outside the USA

• Economic model includes gross domestic product, cross-
exchange rates, arrivals into the USA
• A staff of 35 analysts and 70 field people survey 1 million park
guests, employees, and travel professionals each year

ENMG 605: Operations Management 5

With 20% of Walt Disney World Resort’s customers coming from outside the United States, its economic model includes such variables as gross domestic
product (GDP), cross-exchange rates, and arrivals into the U.S.

Disney also uses 35 analysts and 70 field people to survey 1 million people each year. The surveys, administered to guests at the parks and its 20 hotels, to
employees, and to travel industry professionals, examine future travel plans and experiences at the parks. This helps forecast not only attendance but behavior
at each ride (e.g., how long people will wait, how many times they will ride).

5
 Forecasting at Disney World

• Inputs to the forecasting model include airline specials, Federal

Reserve policies, Wall Street trends, vacation/holiday
schedules for 3,000 school districts around the world
• Average forecast error for the 5-year forecast is 5%
• Average forecast error for annual forecasts is between 0% and
3%

ENMG 605: Operations Management 6

Inputs to the monthly forecasting model include airline specials, speeches by the chair of the Federal Reserve, and Wall Street trends. Disney even monitors
3,000 school districts inside and outside the U.S. for holiday/vacation schedules. With this approach, Disney’s 5-year attendance forecast yields just a 5% error on
average. Its annual forecasts have a 0% to 3% error

6
 What is Forecasting?

• Process of predicting a future event

• Underlying basis
of all business decisions
• Production
• Inventory
• Personnel
• Facilities

ENMG 605: Operations Management 7

Every day, managers like those at Disney make decisions without knowing what will happen in the future. They order inventory without knowing what sales will
be, purchase new equipment despite uncertainty about demand for products, and make investments without knowing what profits will be.

Forecasting is the art and science of predicting future events. Forecasting may involve taking historical data (such as past sales) and projecting them into the
future with a mathematical model.

7
 Forecasting Time Horizons

• Short-range forecast
• Up to 1 year, generally less than 3 months
• Purchasing, job scheduling, workforce levels, job assignments,
production levels
• Medium-range forecast
• 3 months to 3 years
• Sales and production planning, budgeting
• Long-range forecast
• 3+ years
• New product planning, facility location, research and
development
ENMG 605: Operations Management 8

A forecast is usually classified by the future time horizon that it covers. Time horizons fall into three categories:
1. Short-range forecast: This forecast has a time span of up to 1 year but is generally less than 3 months. It is used for planning purchasing, job scheduling,
workforce levels, job assignments, and production levels.
2. Medium-range forecast: A medium-range, or intermediate, forecast generally spans from 3 months to 3 years. It is useful in sales planning, production
planning and budgeting, cash budgeting, and analysis of various operating plans.
3. Long-range forecast: Generally 3 years or more in time span, long-range forecasts are used in planning for new products, capital expenditures, facility
location or expansion, and research and development.

8
 Distinguishing Differences

• Medium/long range forecasts deal with more comprehensive

issues and support management decisions regarding planning
and products, plants and processes
• Short-term forecasting usually employs different
methodologies than longer-term forecasting
• Short-term forecasts tend to be more accurate than longer-
term forecasts

ENMG 605: Operations Management 9

Medium and long-range forecasts are distinguished from short-range forecasts by three features:
1. First, intermediate and long-range forecasts deal with more comprehensive issues supporting management decisions regarding planning and products,
plants, and processes. Implementing some facility decisions, such as GM’s decision to open a new Brazilian manufacturing plant, can take 5 to 8 years from
inception to completion.
2. Second, short-term forecasting usually employs different methodologies than longer-term forecasting. Mathematical techniques, such as moving averages,
exponential smoothing, and trend extrapolation (all of which we shall examine shortly), are common to shortrun projections. Broader, less quantitative
methods are useful in predicting such issues as whether a new product, like the optical disk recorder, should be introduced into a company’s product line.
3. Finally, as you would expect, short-range forecasts tend to be more accurate than longer range forecasts. Factors that influence demand change every day.
Thus, as the time horizon lengthens, it is likely that forecast accuracy will diminish. It almost goes without saying, then, that sales forecasts must be updated
regularly to maintain their value and integrity. After each sales period, forecasts should be reviewed and revised.

9
 Types of Forecasts

• Economic forecasts
• Address business cycle – inflation rate, money supply,
housing starts, etc.
• Technological forecasts
• Predict rate of technological progress
• Impacts development of new products
• Demand forecasts
• Predict sales of existing products and services

ENMG 605: Operations Management 10

Organizations use three major types of forecasts in planning future operations:

1. Economic forecasts address the business cycle by predicting inflation rates, money supplies, housing starts, and other planning indicators.
2. Technological forecasts are concerned with rates of technological progress, which can result in the birth of exciting new products, requiring new plants and
equipment
3. Demand forecasts are projections of demand for a company’s products or services. Forecasts drive decisions, so managers need immediate and accurate
information about real demand

Economic and technological forecasting are specialized techniques that may fall outside the role of the operations manager. The emphasis in this chapter will
therefore be on demand forecasting.

10
 Seven Steps in Forecasting

1. Determine the use of the forecast

2. Select the items to be forecasted
3. Determine the time horizon of the forecast
4. Select the forecasting model(s)
5. Gather the data needed to make the forecast
6. Make the forecast
7. Validate and implement the results

ENMG 605: Operations Management 11

Forecasting follows seven basic steps. We use Disney World, as an example of each step:
1. Determine the use of the forecast: Disney uses park attendance forecasts to drive decisions about staffing, opening times, ride availability, and food
supplies
2. Select the items to be forecasted: For Disney World, there are six main parks. A forecast of daily attendance at each is the main number that determines
labor, maintenance, and scheduling.
3. Determine the time horizon of the forecast: Is it short, medium, or long term? Disney develops daily, weekly, monthly, annual, and 5-year forecasts
4. Select the forecasting model(s): Disney uses a variety of statistical models that we shall discuss, including moving averages, econometrics, and regression
analysis. It also employs judgmental, or nonquantitative, models.
5. Gather the data needed to make the forecast: Disney’s forecasting team employs 35 analysts and 70 field personnel to survey 1 million people/businesses
every year. Disney also uses a firm called Global Insights for travel industry forecasts and gathers data on exchange rates, arrivals into the U.S., airline
specials, Wall Street trends, and school vacation schedules
6. Make the forecast
7. Validate and implement the results: At Disney, forecasts are reviewed daily at the highest levels to make sure that the model, assumptions, and data are
valid. Error measures are applied; then the forecasts are used to schedule personnel down to 15-minute intervals.

These seven steps present a systematic way of initiating, designing, and implementing a forecasting system.

11
 The Realities!

• Forecasts are not perfect, unpredictable outside factors may

impact the forecast
• Most techniques assume an underlying stability in the system
• Product family and aggregated forecasts are more accurate than
individual product forecasts

ENMG 605: Operations Management 12

Regardless of the system that frms like Disney use, each company faces several realities:
• Outside factors that we cannot predict or control often impact the forecast.
• Most forecasting techniques assume that there is some underlying stability in the system. Consequently, some frms automate their predictions using
computerized forecasting software, then closely monitor only the product items whose demand is erratic.
• Both product family and aggregated forecasts are more accurate than individual product forecasts. Disney, for example, aggregates daily attendance
forecasts by park. This approach helps balance the over- and underpredictions for each of the six attractions.

12
 Knowledge Check

1)Forecasts of individual products tend to be more accurate than

forecasts of product families.
True False
2) The forecasting time horizon that would typically be easiest to
predict for would be
a)long-range.
b)medium-range.
c)intermediate range.
d) short-range

ENMG 605: Operations Management 13

13
 Knowledge Check

A forecast that projects a company's sales is a(n)

a)economic forecast.
b)technological forecast.
c)demand forecast.
d)associative model.

ENMG 605: Operations Management 14

14
 Forecasting Approaches

There are two general approaches to forecasting, just as there are two ways to tackle all decision modeling. One is a quantitative analysis; the other is a
qualitative approach. Quantitative forecasts use a variety of mathematical models that rely on historical data and/or associative variables to forecast demand.
Subjective or qualitative forecasts incorporate such factors as the decision maker’s intuition, emotions, personal experiences, and value system in reaching a
forecast. Some firms use one approach and some use the other. In practice, a combination of the two is usually most effective.

15
 Qualitative Methods

• Used when situation is vague and little data exist

• New products
• New technology
• Involves intuition, experience
• e.g., forecasting sales on Internet

ENMG 605: Operations Management 16

16
 Overview of Qualitative Methods

1. Jury of executive opinion

– Pool opinions of high-level experts, sometimes
augmented by statistical models
2. Delphi method
– Panel of experts, queried iteratively
3. Sales force composite
– Estimates from individual salespersons are reviewed for
reasonableness, then aggregated
4. Market Survey
– Ask the customer
ENMG 605: Operations Management 17

1. Jury of executive opinion: Under this method, the opinions of a group of high-level experts or managers, often in combination with statistical models, are
pooled to arrive at a group estimate of demand
2. Delphi method: There are three different types of participants in the Delphi method: decision makers, staff personnel, and respondents. Decision makers
usually consist of a group of 5 to 10 experts who will be making the actual forecast. Staff personnel assist decision makers by preparing, distributing,
collecting, and summarizing a series of questionnaires and survey results. The respondents are a group of people, often located in different places, whose
judgments are valued. This group provides inputs to the decision makers before the forecast is made.
3. Sales force composite: In this approach, each salesperson estimates what sales will be in his or her region. These forecasts are then reviewed to ensure
that they are realistic. Then they are combined at the district and national levels to reach an overall forecast.
4. Market survey: This method solicits input from customers or potential customers regarding future purchasing plans.

17
 Quantitative method

• Used when situation is ‘stable’ and historical data exist

• Existing products
• Current technology
• Involves mathematical techniques
• e.g., forecasting sales of color televisions

ENMG 605: Operations Management 18

18
 Overview of Quantitative Methods

1. Naive approach
2. Moving averages Time-series models
3. Exponential smoothing
4. Trend projection
5. Linear regression Associative model

ENMG 605: Operations Management 19

Five quantitative forecasting methods, all of which use historical data, are described in this chapter. They fall into two categories: Time-series models and
Associative model

19
Time Series Forecasting

20
 Time Series Models

• Set of evenly spaced numerical data

• Obtained by observing response variable at regular time
periods

• Forecast based only on past values, no other variables

important
• Assumes that factors influencing past and present will
continue influence in future
• Time series methods only uses past values of the
phenomena

ENMG 605: Operations Management 21

Time-series models predict on the assumption that the future is a function of the past. In other words, they look at what has happened over a period of time and
use a series of past data to make a forecast. If we are predicting sales of lawn mowers for example, we only use the past sales for lawn mowers as input to make
the forecasts.

21
 Time-Series Components

ENMG 605: Operations Management 22

Analyzing time series means breaking down past data into components and then projecting them forward. A time series has four components:
1. Trend is the gradual upward or downward movement of the data over time.
2. Seasonality is a data pattern that repeats itself after a period of days, weeks, months, or quarters.
3. Cycles are patterns in the data that occur every several years.
4. Random variations are “blips” in the data caused by chance and unusual situations. They follow no discernible pattern, so they cannot be predicted.

22
 Components of Demand

ENMG 605: Operations Management 23

fffffFigure 4.1 illustrates a demand over a 4-year period. It shows the average, trend, seasonal components, and random variations around the demand curve.
The average demand is the sum of the demand for each period divided by the number of data periods.

23
 Trend Component

• Persistent, overall upward or downward pattern

• Changes due to population, technology, age, culture, etc.
• Typically several years duration

ENMG 605: Operations Management 24

Trend is the gradual upward or downward movement of the data over time. Changes in income, population, age distribution, or cultural views may account for
movement in trend.

24
 Seasonal Component

• Regular pattern of up and down fluctuations

• Due to weather, customs, etc.
• Occurs within a single year

ENMG 605: Operations Management 25

Seasonality is a data pattern that repeats itself after a period of days, weeks, months, or quarters. There are six common seasonality patterns

25
 Cyclical Component

• Repeating up and down movements

• Affected by business cycle, political, and economic factors
• Multiple years duration
• Often causal or associative relationships

ENMG 605: Operations Management 26

Cycles are patterns in the data that occur every several years. They are usually tied into the business cycle and are of major importance in short-term business
analysis and planning. Predicting business cycles is difficult because they may be affected by political events or by international turmoil

26
 Random Component

• Erratic, unsystematic, ‘residual’ fluctuations

• Due to random variation or unforeseen events
• Short duration and nonrepeating

ENMG 605: Operations Management 27

Random variations are “blips” in the data caused by chance and unusual situations. They follow no discernible pattern, so they cannot be predicted.

27
 Stationary Series

A stationary series 𝐷 satisfies the following:

𝐷 =𝜇+𝜖
𝐸 𝜖 = 0, 𝑣𝑎𝑟 𝜖 = 𝜎

ENMG 605: Operations Management 28

A stationary series is a time series in which the demand fluctuates around a constant value. In particular the series is assumed to be the sum of a constant and
random variable with a 0 mean.

28
 Knowledge Check

Time-series analysis is based on the assumption that

a. random error terms are normally distributed.
b. there are dependable correlations between the variable to be
forecast and other independent variables.
c. past patterns in the variable to be forecasted will continue
unchanged into the future.
d. the data do not exhibit a trend

ENMG 605: Operations Management 29

29
 Knowledge Check

Which of the following can’t be a component for a time series plot?

A) Seasonality
B) Trend
C) Cyclical
D) Noise
E) None of the above

ENMG 605: Operations Management 30

30
 Knowledge Check

The following data exhibits:

a) Seasonality b) Trend c) none of the
above d) all of the above
ENMG 605: Operations Management 31

31
 Knowledge Check

Time series patterns that repeat themselves after a period of days

or weeks are called
seasonality.
cycles.
random variation.
trend

ENMG 605: Operations Management 32

32
Methods for Stationary Series

33
 Naive Approach

• Assumes demand in next period is the same as demand in

most recent period
• e.g., If January sales were 68, then February sales will
be 68
• Sometimes cost effective and efficient
• Can be good starting point

ENMG 605: Operations Management 34

The simplest way to forecast is to assume that demand in the next period will be equal to demand in the most recent period. In other words, if sales of a
product—say, Nokia cell phones—were 68 units in January, we can forecast that February’s sales will also be 68 phones.
In some product lines, this naive approach is the most cost-effective and efficient objective forecasting model. At least it provides a starting point against which
more sophisticated models that follow can be compared.

34
 Moving Averages

• MA is a series of arithmetic means

• Used if little or no trend
• Used often for smoothing
• Provides overall impression of data over time
• 𝑀𝐴 𝑁 : 𝐹 = 1/𝑁 ∑ 𝐷 = (1/𝑁)(𝐷 +⋯+𝐷 )
• 𝐹 = 1/𝑁 ∑ 𝐷 = 1/𝑁 𝐷 + ∑ 𝐷 −𝐷 =
𝐹 + 1/𝑁 𝐷 − 𝐷
• This means we only need to compute 𝐷_𝑡−𝐷_(𝑡−𝑁) to update
the forecast.

ENMG 605: Operations Management 35

A moving-average forecast uses a number of historical actual data values to generate a forecast. Moving averages are useful if we can assume that market
demands will stay fairly steady over time. A 4-month moving average is found by simply summing the demand during the past 4 months and dividing by 4. With
each passing month, the most recent month’s data are added to the sum of the previous 3 months’ data, and the earliest month is dropped. This practice tends
to smooth out short-term irregularities in the data series.

35
 Moving Average Example

ENMG 605: Operations Management 36

Donna’s Garden Supply wants a 3-month moving-average forecast, including a forecast for next January, for shed sales.
To project the demand for sheds in the coming January, we sum the October, November, and December sales and divide by 3: January forecast = (18 + 16 + 14)/3
= 16

36
 Moving Average lags behind the trend

• Consider a demand process

in which there is a definite
trend. For example, suppose
that
• the observed demand is 2,
4, 6, 8, 10, 12, 14, 16, 18,
20, 22, 24. Consider the
one-step ahead MA(3) and
MA(6) forecasts for this
series.

ENMG 605: Operations Management 37

In t

37
 Moving Average lags behind the trend

ENMG 605: Operations Management 38

Notice that both the MA(3) and the MA(6) forecasts lag behind the trend. Furthermore, MA(6) has a greater lag. This implies that the use of simple moving
averages is not an appropriate forecasting method when there is a trend in the series.

38
 Weighted Moving Average

• Used when some trend might be present

• Older data usually less important
• Weights based on experience and intuition
• 𝑊𝑀𝐴 𝑁 : 𝐹 = ∑ 𝑤 𝐷 /∑ 𝑤

ENMG 605: Operations Management 39

When a detectable trend or pattern is present, weights can be used to place more emphasis on recent values. This practice makes forecasting techniques more
responsive to changes because more recent periods may be more heavily weighted. Choice of weights is somewhat arbitrary because there is no set formula to
determine them. Therefore, deciding which weights to use requires some experience.

39
 Weighted Moving Average Example

ENMG 605: Operations Management 40

Donna’s Garden Supply (see Example 1) wants to forecast storage shed sales by weighting the past 3 months, with more weight given to recent data to make
them more significant

40
 Weighted Moving Average Example

ENMG 605: Operations Management 41

The results of this weighted-average forecast are as follows:

41
 Weighted Moving Average Example

ENMG 605: Operations Management 42

Figure 4.2, a plot of the data in Examples 1 and 2, illustrates the lag effect of the moving average models. Note that both the moving-average and weighted-
moving-average lines lag the actual demand. The weighted moving average, however, usually reacts more quickly to demand changes. Even in periods of
downturn (see November and December), it more closely tracks the demand.

42
 Potential Problems With Moving Average

• Increasing n smooths the forecast but makes it less sensitive

to changes
• Does not forecast trends well
• Requires extensive historical data

ENMG 605: Operations Management 43

Both simple and weighted moving averages are effective in smoothing out sudden fluctuations in the demand pattern to provide stable estimates. Moving
averages do, however, present three problems:
1. Increasing the size of n (the number of periods averaged) does smooth out fluctuations better, but it makes the method less sensitive to changes in the
data.
2. Moving averages cannot pick up trends very well. Because they are averages, they will always stay within past levels and will not predict changes to either
higher or lower levels. That is, they lag the actual values.
3. Moving averages require extensive records of past data.

43
 Exponential Smoothing

• Form of weighted moving average

• Weights decline exponentially
• Most recent data weighted most
• Requires smoothing constant (𝛼)
• Ranges from 0 to 1
• Subjectively chosen
• Involves little record keeping of past data

ENMG 605: Operations Management 44

Exponential smoothing is another weighted-moving-average forecasting method. It involves very little record keeping of past data and is fairly easy to use.

44
 Exponential Smoothing

New forecast =Last period’s forecast+ a (Last period’s actual

demand – Last period’s forecast)
Ft = Ft – 1 + a(Dt – 1 – Ft – 1)

where Ft = new forecast

Ft – 1 = previous period’s forecast
a = smoothing (or weighting) constant (0 ≤ a ≤ 1)
Dt – 1 = previous period’s actual demand

ENMG 605: Operations Management 45

alpha is a weight, or smoothing constant, chosen by the forecaster. The concept is not complex. The latest estimate of demand is equal to the old forecast
adjusted by a fraction of the difference between the last period’s actual demand and last period’s forecast.

45
 Exponential Smoothing

• 𝐹 = 𝛼𝐷 + 1 − 𝛼 𝐹 = 𝛼𝐷 + 1 − 𝛼 (𝛼𝐷 +
1 − 𝛼 𝐹 ) = 𝛼𝐷 + 𝛼 1 − 𝛼 𝐷 + 1 − 𝛼 𝐹
• In general 𝐹 = 𝛼𝐷 + 𝛼 1 − 𝛼 𝐷 + 𝛼 1 − 𝛼 𝐷 + ⋯

ENMG 605: Operations Management 46

MA gives weight to all past data which is exponen ally decreasing with the data age at rate (1−a).

46
 Exponential Smoothing Example

In January, a car dealer predicted February demand for 142 Ford

Mustangs. Actual February demand was 153 autos. Using a
smoothing constant chosen by management of 𝑎 = .20, the dealer
wants to forecast March demand using the exponential smoothing
model.

March forecast = 142 + .2(153 – 142) = 144.2 ≈ 144 cars

ENMG 605: Operations Management 47

47
 Effect of Smoothing Constants

• Smoothing constant generally .05 ≤ a ≤ .50

• As a increases, older values become less significant

ENMG 605: Operations Management 48

The smoothing constant, alpha, is generally in the range from .05 to .50 for business applications. It can be changed to give more weight to recent data (when a is
high) or more weight to past data (when alpha is low). When alpha reaches the extreme of 1.0 All the older values drop out, and the forecast becomes identical
to the naive model mentioned earlier in this chapter

The following table helps illustrate this concept. For example, when alpha = .5, we can see that the new forecast is based almost entirely on demand in the last
three or four periods. When a = .1, the forecast places little weight on recent demand and takes many periods (about 19) of historical values into account

48
 Impact of Different a

• Choose high
values of 𝛼 when
underlying average
is likely to change
• Choose low values
of a when
underlying average
is stable

ENMG 605: Operations Management 49

As 𝛼 increases, the predicted value exhibits greater variation.

49
 Note on Exponential Smoothing

• Because Exponential Smoothing requires that each stage we

have the previous forecast, it is not obvious how to get the
method started.
• We could assume that the initial forecast is equal to the initial
value of demand. However this approach has a serious
drawback. Exponential smoothing puts substantial weight on
past observations.
• This problem can be overcome by allowing the process to
evolve for a reasonable number of periods(say 10 or more)
and using the average of the demand during those periods as
the initial forecast.
ENMG 605: Operations Management 50

50
 Knowledge Check

If demand is 106 during January, 120 in February, 134 in March,

and 142 in April, what is the 3-month simple moving average for
May?
a)132
b)126
c)142
d)138

ENMG 605: Operations Management 51

51
 Knowledge Check

Given last period's forecast of 65, and last period's demand of 62,
what is the simple exponential smoothing forecast with an alpha of
0.4 for the next period?
62
63.2
65
63.8

ENMG 605: Operations Management 52

52
 Knowledge Check

Sum of weights in exponential smoothing is

A) <1
B) 1
C) >1
D) None of the above
Which of the following smoothing constants would make an
exponential smoothing forecast equivalent to a naive forecast?
A) 0 B).01 C).1 D).5 E) 1.0

ENMG 605: Operations Management 53

53
 Selecting the Smoothing Constant

The objective is to obtain the most accurate forecast no matter the

technique
We generally do this by selecting the model that gives us the lowest
forecast error.

ENMG 605: Operations Management 54

The appropriate value of the smoothing constant, alpha , can make the difference between an accurate forecast and an inaccurate forecast. High values of alpha
are chosen when the underlying average is likely to change. Low values of a are used when the underlying average is fairly stable. In picking a value for the
smoothing constant, the objective is to obtain the most accurate forecast.

54
 Measuring Forecast Error

• The overall accuracy of any forecasting model—moving average,

exponential smoothing, or other—can be determined by
comparing the forecasted values with the actual or observed
values.
• Forecast error = Actual demand - Forecast value
𝑒 =𝐷 − 𝐹

ENMG 605: Operations Management 55

55
 Common Measures of Error

Mean Absolute Deviation (MAD)

∑ |𝑒 |
𝑀𝐴𝐷 =
𝑛
Mean squared error (MSE)
∑ 𝑒
𝑀𝑆𝐸 =
𝑛

Mean absolute percentage error (MAPE)

∑ |𝑒 /𝐷 |
𝑀𝐴𝑃𝐸 = 100
𝑛

ENMG 605: Operations Management 56

Several measures are used in practice to calculate the overall forecast error. These measures can be used to compare different forecasting models, as well as to
monitor forecasts to ensure they are performing well. Three of the most popular measures are mean absolute deviation (MAD), mean squared error (MSE), and
mean absolute percent error (MAPE).

56
 Common Measures of Error

MEASURE MEANING
Mean absolute How much the forecast
deviation (MAD) missed the target
Mean squared The square of how much
error (MSE) the forecast missed the
target
Mean absolute The average percent
percent error error
(MAPE)

ENMG 605: Operations Management 57

A drawback of using the MSE is that it tends to accentuate large deviations due to the squared term. Hence, using MSE as the measure of forecast error typically
indicates that we prefer to have several smaller deviations rather than even one large deviation

A problem with both the MAD and MSE is that their values depend on the magnitude of the item being forecast. If the forecast item is measured in thousands,
the MAD and MSE values can be very large.
The MAPE is perhaps the easiest measure to interpret since its output is a percentage independent on the magnitude of the input data

57
 Comparison of Forecast Error

ENMG 605: Operations Management 58

Consider the example earlier. Two values of alpha will be compared.

58
 MAD(a = .10) = 82.45/8 = 10.31
MAD(a = .50)== 98.62/8 = 12.33
MSE(a = .10) =1,526.52/8 = 190.8
MSE(a = .50)= 1,561.91/8 = 195.24
MAPE(a = .10)= 44.75%/8 = 5.59%
MAPE(a = .50) = 54.00%/8 = 6.75%

ENMG 605: Operations Management 59

59
 ENMG 605: Operations Management 60

Alpha=0.1 yields the better forecast.

60
 Knowledge Check

For this set of errors: -1, -4, 0, +2, +3, MAD is:
A. 1.0
B. 1.6
C. 2.0
D. 2.5
E. 10.0

ENMG 605: Operations Management 61

61
 Knowledge Check

The mean absolute deviation error (MAD) will always be:

positive.
between 0 and 1.
negative.
None of the above

ENMG 605: Operations Management 62

62
Methods for non-Stationary Series

63
 Exponential Smoothing with Trend Adjustment

Simple exponential smoothing, is like any other moving-average

technique fails to respond to trends and lags behind it

ENMG 605: Operations Management 64

Assume that demand for our product or service has been increasing by 100 units per month and that we have been forecasting with alpha = 0.4 in our
exponential smoothing model. The following table shows a severe lag in the second, third, fourth, and fifth months, even when our initial estimate for month 1 is
perfect

64
 Exponential Smoothing with Trend Adjustment

• Also called Double Exponential Smoothing

• Applies when demand data has a trend with random noise.
• Estimates the trend line.

ENMG 605: Operations Management 65

65
 Exponential Smoothing with Trend Adjustment

Forecast Exponentially Exponentially

including (FITt) = smoothed (Ft) + smoothed (Tt)
trend forecast trend
Ft = a(Dt - 1) + (1 - a)(Ft - 1 + Tt - 1)
Tt = b(Ft - Ft - 1) + (1 - b)Tt - 1
Ft = exponentially smoothed forecast average
Tt = exponentially smoothed trend
Dt = actual demand
a = smoothing constant for average (0 ≤ a ≤ 1)
b = smoothing constant for trend (0 ≤ b ≤ 1)

ENMG 605: Operations Management 66

To improve our forecast, let us illustrate a more complex exponential smoothing model, one that adjusts for trend. The idea is to compute an exponentially
smoothed average of the data and then adjust for positive or negative lag in trend. With trend-adjusted exponential smoothing, estimates for both the average
and the trend are smoothed. This procedure requires two smoothing constants: alpha for the average and beta for the trend.

66
 Exponential Smoothing with Trend Adjustment

Step 1: Compute Ft
Step 2: Compute Tt
Step 3: Calculate the forecast FITt = Ft + Tt

Ft = a(Dt - 1) + (1 - a)(Ft - 1 + Tt - 1)
Tt = b(Ft - Ft - 1) + (1 - b)Tt - 1

ENMG 605: Operations Management 67

The three steps to compute a trend-adjusted forecast are

67
 Exponential Smoothing with Trend Adjustment Example

A large Portland manufacturer wants to forecast demand for a piece of pollution-control equipment. A
review of past sales, as shown below, indicates that an increasing trend is present

ENMG 605: Operations Management 68

A large Portland manufacturer wants to forecast demand for a piece of pollution-control equipment. A review of past sales, as shown below, indicates that an
increasing trend is present

68
 Exponential Smoothing with Trend Adjustment Example

Smoothing constants are assigned the values of 𝑎 = .2 and 𝛽 =

.4. The firm assumes the initial forecast average for month 1 (𝐹 )
was 11 units and the trend over that period (𝑇 ) was 2 units

ENMG 605: Operations Management 69

69
 Exponential Smoothing with Trend Adjustment Example

Step 1: Average for Month 2

F2 = aA1 + (1 – a)(F1 + T1)
F2 = (.2)(12) + (1 – .2)(11 + 2)
= 2.4 + (.8)(13) = 2.4 + 10.4
= 12.8 units

ENMG 605: Operations Management 70

70
 Exponential Smoothing with Trend Adjustment Example

Step 2: Trend for Month 2

T2 = b(F2 - F1) + (1 - b)T1
T2 = (.4)(12.8 - 11) + (1 - .4)(2)
= .72 + 1.2 = 1.92 units

ENMG 605: Operations Management 71

71
 Exponential Smoothing with Trend Adjustment Example

Step 3: Calculate FIT for Month 2

FIT2 = F2 + T 2
FIT2 = 12.8 + 1.92
= 14.72 units

ENMG 605: Operations Management 72

72
 Exponential Smoothing with Trend Adjustment Example

ENMG 605: Operations Management 73

73
 Exponential Smoothing with Trend Adjustment Example

ENMG 605: Operations Management 74

Double exponential smoothing is more responsive to the trend than the original exponential smoothing.

74
Trend Projections

75
 Trend Projections

• Fitting a trend line to historical data points to project into the

medium to long-range
• Linear trends can be found using the least squares technique
• 𝑦 = 𝑎 + 𝑏𝑥
where 𝑦 = computed value of the variable to be predicted
(dependent variable)
a = y-axis intercept
b = slope of the regression line
x = the independent variable

ENMG 605: Operations Management 76

The last time-series forecasting method we will discuss is trend projection. This technique fits a trend line to a series of historical data points and then projects
the slope of the line into the future for medium to long-range forecasts

76
 Least Squares Method

ENMG 605: Operations Management 77

Statisticians have developed equations that we can use to find the values of a and b for any regression line. The Least Squares Method results in a straight line
that minimizes the sum of the squares of the vertical differences or deviations from the line to each of the actual observations. Figure 4.4 illustrates the least-
squares approach.

77
 Least Squares Method

Equations to calculate the regression variables

ŷ = a + bx

b=
å xy - nxy
å x - nx 2 2

a = y - bx
ENMG 605: Operations Management 78

78
 Sample Correlation Coefficient R

The correlation coefficient R is:

∑ (𝑥 − 𝑋 )(𝑥 −𝑌 )
𝑅=
∑ (𝑥 − 𝑋 ) ∑ (𝑦 − 𝑌 )

It is an indication on how good the fit is.

ENMG 605: Operations Management 79

79
 Sample Correlation Coefficient R

• Measures the linear association between y and x.

• The correlation coefficient is “unitless” and ranges between -1
and +1.

ENMG 605: Operations Management 80

80
 Sample Correlation Coefficient R

ENMG 605: Operations Management 81

81
 Least Squares Example

• The demand for electric power at N.Y. Edison over the past 7
years is shown in the following table, in megawatts. The firm
wants to forecast next year’s demand by fitting a straight-line
trend to these data.

ENMG 605: Operations Management 82

82
 Least Squares Example

b=
å xy - nxy = 3,063 - (7) ( 4) (98.86) = 295 = 10.54
å x - nx 140 - ( 7) ( 4 )
2 2 2
28

()
a = y - bx = 98.86 -10.54 4 = 56.70
Thus, ŷ = 56.70 +10.54x
Demand in year 8 = 56.70 + 10.54(8)
= 141.02, or 141 megawatts

ENMG 605: Operations Management 83

83
 Least Squares Example on Excel
*Screen Recording*

ENMG 605: Operations Management 84

The analysis can also be done on excel. R^2=0.8009 which means the line is a good fit.

84
 Least Squares Requirements

• We always plot the data to insure a linear relationship

• Check the value of 𝑅 to see of the trend is a good fit
• We do not predict time periods far beyond the database
• Deviations around the least squares line are assumed to be
random

ENMG 605: Operations Management 85

Using the least-squares method implies that we have met four requirements:
1. We always plot the data because least-squares data assume a linear relationship. If a curve appears to be present, curvilinear analysis is probably needed.
2. Check the value of R^2, if the value is less than 0.7 it means the trend is not a good fit
3. We do not predict time periods far beyond our given database. For example, if we have 20 months’ worth of average prices of Microsoft stock, we can
forecast only 3 or 4 months into the future. Forecasts beyond that have little statistical validity. Thus, you cannot take 5 years’ worth of sales data and
project 10 years into the future. The world is too uncertain.
4. Deviations around the least-squares line are assumed to be random and normally distributed, with most observations close to the line and only a smaller
number farther out.

85
 Knowledge Check

The correlation coefficient is used to determine:

a. A specific value of the y-variable given a specific value of the x-
variable
b. A specific value of the x-variable given a specific value of the y-
variable
c. The strength of the relationship between the x and y variables
d. None of these

ENMG 605: Operations Management 86

86
 Knowledge Check

Larger values of 𝑅 imply that the observations are more closely

grouped about the
a. average value of the time
b. average value of the forecasted value
c. least squares line
d. origin

ENMG 605: Operations Management 87

87
 Knowledge Check

A fitted least squares regression line

a. may be used to predict a value of y if the corresponding x value
is given
b. is evidence for a cause-effect relationship between x and y
c. can only be computed if a strong linear relationship exists
between x and y
d. None of these alternatives is correct.

ENMG 605: Operations Management 88

88
Seasonal Variations In Data

89
 Seasonal Variations In Data

• Demand for many products is seasonal. Yamaha, the manufacturer of this jet ski
and snowmobile, produces products with complementary demands to address
seasonal fluctuations
• The multiplicative seasonal model can adjust trend data for seasonal variations in
demand

ENMG 605: Operations Management 90

Seasonal variations in data are regular movements in a time series that relate to recurring events such as weather or holidays. Demand for coal and fuel oil, for
example, peaks during cold winter months. Demand for golf clubs or sunscreen may be highest in summer. Seasonality may be applied to hourly, daily, weekly,
monthly, or other recurring patterns.

90
 Seasonal Variations In Data

Steps in the process for monthly seasons:

1. Find average historical demand for each month
2. Compute the average demand over all months
3. Compute a seasonal index for each month
4. Estimate next year’s total demand
5. Divide this estimate of total demand by the number of months,
then multiply it by the seasonal index for that month

ENMG 605: Operations Management 91

In what is called a multiplicative seasonal model, seasonal factors are multiplied by an estimate of average demand to produce a seasonal forecast. Our
assumption in this section is that trend has been removed from the data. Otherwise, the magnitude of the seasonal data will be distorted by the trend. Here are
the steps we will follow for a company that has “seasons” of 1 month

1. Find the average historical demand each season (or month in this case) by summing the demand for that month in each year and dividing by the number of
years of data available. For example, if, in January, we have seen sales of 8, 6, and 10 over the past 3 years, average January demand equals (8 + 6 + 10)/3 =
8 units.
2. Compute the average demand over all months by dividing the total average annual demand by the number of seasons. For example, if the total average
demand for a year is 120 units and there are 12 seasons (each month), the average monthly demand is 120/12 = 10 units.
3. Compute a seasonal index for each season by dividing that month’s historical average demand (from Step 1) by the average demand over all months (from
Step 2). For example, if the average historical January demand over the past 3 years is 8 units and the average demand over all months is 10 units, the
seasonal index for January is 8/10 = .80. Likewise, a seasonal index of 1.20 for February would mean that February’s demand is 20% larger than the average
demand over all months.
4. Estimate next year’s total annual demand.
5. Divide this estimate of total annual demand by the number of seasons, then multiply it by the seasonal index for each month. This provides the seasonal
forecast

91
 Seasonal Index Example

ENMG 605: Operations Management 92

92
 Seasonal Index Example

Average
1,128
monthly = = 94
demand 12 months

ENMG 605: Operations Management 93

93
 Seasonal Index Example

Seasonal = Average monthly demand for past 3 years

index Average monthly demand

ENMG 605: Operations Management 94

94
 Seasonal Index Example

If we expect the annual demand for computers to be 1,200 units next year, we would use these seasonal
indices to forecast the monthly demand as follows:

ENMG 605: Operations Management 95

95
 Seasonal Index Example

• Seasonal forecast for Year 4

ENMG 605: Operations Management 96

Think of these indices as percentages of average sales. The average sales (without seasonality) would be 94, but with seasonality, sales fluctuate from 85% to
131% of average.

96
 Seasonal Index Example

ENMG 605: Operations Management 97

Notice how the pattern is similar for every year

97
APPLYING BOTH TREND AND SEASONAL INDICES

98
 APPLYING BOTH TREND AND SEASONAL INDICES

• In some cases, both trend and seasonality exist in the data

• For a better forecast, both trend and seasonality should be taken
in consideration

ENMG 605: Operations Management 99

Notice how the pattern is similar for every year

99
 APPLYING BOTH TREND AND SEASONAL INDICES

• San Diego Hospital wants to improve its forecasting by

applying both trend and seasonal indices to 66 months of data
it has collected. It will then forecast “patient-days” over the
coming year
• Using 66 months of adult inpatient hospital days, the trend line
was computed: 𝑦 = 8,090 + 21.5𝑥
Where 𝑦 = patient days, 𝑥 = time, in months
• Based on this model, which reflects only trend data, the
hospital forecasts patient days for the next month (period 67)
to be:
Patient days = 8,090 + (21.5)(67) = 9,530 (trend only)
ENMG 605: Operations Management 100

100
 APPLYING BOTH TREND AND SEASONAL INDICES

• While this model, as plotted in, recognized the upward trend

line in the demand for inpatient services, it ignored the
seasonality that the administration knew to be present.

ENMG 605: Operations Management 101

101
 APPLYING BOTH TREND AND SEASONAL INDICES

• The following table provides seasonal indices based on the

same 66 months. Such seasonal data, by the way, were found
to be typical of hospitals nationwide.

ENMG 605: Operations Management 102

102
 APPLYING BOTH TREND AND SEASONAL INDICES

ENMG 605: Operations Management 103

These seasonal indices are graphed. Note that January, March, July, and August seem to exhibit significantly higher patient days on average, while February,
September, November, and December experience lower patient days. However, neither the trend data nor the seasonal data alone provide a reasonable
forecast for the hospital

103
 APPLYING BOTH TREND AND SEASONAL INDICES

Patient days = (Trend-adjusted forecast)(Monthly seasonal index) = (9,530)(1.04) = 9,911

ENMG 605: Operations Management 104

Only when the hospital multiplied the trend-adjusted data by the appropriate seasonal index did it obtain good forecasts. Thus, for period 67 (January):

Patient days = (Trend-adjusted forecast)(Monthly seasonal index) = (9,530)(1.04) = 9,911

104
 APPLYING BOTH TREND AND SEASONAL INDICES

ENMG 605: Operations Management 105

A graph showing the forecast that combines both trend and seasonality is shown.

With trend only, the September forecast is 9,702, but with both trend and seasonal adjustments, the forecast is 9,411. By combining trend and seasonal data, the
hospital was better able to forecast inpatient days and the related staffing and budgeting vital to effective operations

105
 Example
*Screen Recording*

Below you are given information on crime statistics for Middletown.

Year Quarter Number of Crimes Comitted
1 1 10
2 20
3 25
4 5
2 1 10
2 30
3 35
4 25
3 1 20
2 40
3 35
4 15
4 1 20
2 50
3 45
4 35

ENMG 605: Operations Management 106

106
 Example
*Screen Recording*

a) Find the trend in the data (use excel)

b) Remove the trend from the data
c) Find the seasonal factors
d) Use the seasonal and trend components to forecast the number
of crimes for each quarter of Year 5.

ENMG 605: Operations Management 107

107
 Example
*Screen Recording*

a) Find the trend in the data (use excel)

Crimes Committed
60 y = 1.7353x + 11.5
R² = 0.3976
50

40

30

20

10

ENMG 605: Operations Management 108

108
 Example
*Screen Recording*

b) Remove the trend from the data

ENMG 605: Operations Management 109

109
 Example
*Screen Recording*

ENMG 605: Operations Management 110

110
 Example
*Screen Recording*

Find the seasonal factors

ENMG 605: Operations Management 111

111
 Example
*Screen Recording*

Forecast the number of

crimes for each quarter
of Year 5.

ENMG 605: Operations Management 112

112
Associative Forecasting

113
 Associative Forecasting

• Forecasting an outcome based on predictor variables using the

least squares technique
^
y = a + bx

where y = value of the dependent variable (in our example, sales)

a = y-axis intercept
b = slope of the regression line
x = the independent variable

ENMG 605: Operations Management 114

Unlike time-series forecasting, associative forecasting models usually consider several variables that are related to the quantity being predicted. Once these
related variables have been found, a statistical model is built and used to forecast the item of interest. This approach is more powerful than the time-series
methods that use only the historical values for the forecast variable.

We can use the same mathematical model that we employed in the least-squares method of trend projection to perform a linear-regression analysis. The
independent variable, x, need no longer be time.

114
 Associative Forecasting Example

• Nodel Construction Company renovates old homes in West Bloomfield,

Michigan. Over time, the company has found that its dollar volume of renovation
work is dependent on the West Bloomfield area payroll. Management wants to
establish a mathematical relationship to help predict sales
• Nodel’s VP of operations has prepared the following table, which lists company
revenues and the amount of money earned by wage earners in West Bloomfield
during the past 6 years:

ENMG 605: Operations Management 115

115
 Associative Forecasting Example

4.0 – ŷ = 1.75 + .25x

Sales = 1.75 + .25(payroll)

Nodel’s sales
(in $ millions)
3.0 –

2.0 – 𝑅 = 0.901, 𝑅 = 0.81

1.0 –

| | | | | | |

0 1 2 3 4 5 6 7
Area payroll (in $ billions)

ENMG 605: Operations Management 116

116
 Knowledge Check

To plan its upcoming programming, a local television station wanted to study the
television watching habits of people in the area. Station executives called a random
sample of people in their broadcast area and asked them a variety of questions. In
particular, they asked people their ages, x, and how many hours of television they had
watched last week, y. Find the equation for the linear regression model that fits the
data
Age Hours
11 4
14 5
20 5
47 12
75 11

ENMG 605: Operations Management 117

117
 •y = 0.124x + 3.275
•y = 2.1x + 1.1
•x = 0.124y + 3.275
•x = 3.275y + 0.124

ENMG 605: Operations Management 118

118
Monitoring and Controlling Forecasts

119
 Monitoring and Controlling Forecasts

Tracking Signal
• Measures how well the forecast is predicting actual values
• Ratio of cumulative forecast errors to mean absolute deviation
(MAD)
• Good tracking signal has low values
• If forecasts are continually high or low, the forecast has a
bias error

ENMG 605: Operations Management 120

One way to monitor forecasts to ensure that they are performing well is to use a tracking signal. A tracking signal is a measurement of how well a forecast is
predicting actual values. As forecasts are updated every week, month, or quarter, the newly available demand data are compared to the forecast values.

120
 Monitoring and Controlling Forecasts

Tracking Cumulative error

signal =
MAD

=
å (Actual demand in period i -Forecast demand in period i)
å Actual -Forecast
n

ENMG 605: Operations Management 121

The tracking signal is computed as the cumulative error divided by the mean absolute deviation (MAD)

121
 Monitoring and Controlling Forecasts

ENMG 605: Operations Management 122

Once tracking signals are calculated, they are compared with predetermined control limits. When a tracking signal exceeds an upper or lower limit, there is a
problem with the forecasting method, and management may want to reevaluate the way it forecasts demand. The Figure shows the graph of a tracking signal
that is exceeding the range of acceptable variation. If the model being used is exponential smoothing, perhaps the smoothing constant needs to be readjusted.

122
 Tracking signal Example

• Develop a tracking signal for the forecast and see if it stays

within acceptable limits, which we define as 4 MADs

ENMG 605: Operations Management 123

123
 Tracking signal Example

ENMG 605: Operations Management 124

Because the tracking signal drifted from -2 MAD to +2.5 MAD we can conclude that it is within acceptable limits.

124
 Knowledge Check

A forecasting technique consistently produces a negative tracking

signal. This means that:
the MSE will also consistently be negative.
the forecasting technique consistently under predicts.
the forecast technique consistently over predicts.
the MAPE will also consistently be negative.

ENMG 605: Operations Management 125

125