MGT321- MGT311

CH10

10.1 FORECASTING FOR DECISION MAKING Although there are many types of forecasting, this chapter will focus on forecasting demand for output from the operations function. Demand and sales, however, are not always the same thing. Whenever the demand is not constrained by capacity or other management policies, the forecasting of demand will be the same as the forecasting of sales. Otherwise, sales may be somewhat below real customer demand. We should also clarify the difference between forecasting and planning. Forecasting deals with what we think will happen in the future. Planning deals with what we think should happen in the future. Thus, through planning, we consciously attempt to alter future events, while we use forecasting only to predict them. Good planning utilizes a forecast as an input. If the forecast is not acceptable, sometimes a plan can be devised to change the course of events.

Forecasting is one input to all types of business planning and control, both inside and outside the operations function. Marketing uses forecasts for planning products and services, promotion, and pricing. Finance uses forecasting as an input to financial planning. Human resources requires forecasts to anticipate hiring decisions and personnel budgets. Forecasting is an input for operations decisions on process design, capacity planning, and inventory. Forecasting is done by firms all along the supply chain. For process design purposes, forecasting is needed to decide on the type of process and the degree of automation to be used. For example, a low forecast of future demand may indicate that little automation is needed and the process should be kept as simple as possible. Capacity decisions utilize forecasts for different planning horizons. For planning facilities, a long-range forecast several years into the future is needed. For medium- range capacity decisions, a more detailed forecast by product line or service will be needed to determine hiring plans, subcontracting, and equipment decisions. Short- range capacity decisions, including the assignment of available workers and machines to jobs require a highly accurate forecast. Operations, Marketing, Finance, and Human Resources collaborate to both create and use

forecasts.© mediaphotos/Getty Images

Inventory decisions resulting in purchasing actions tend to be short range in nature and deal with specific products. The forecasts that lead to these decisions must meet the same requirements as short-range scheduling forecasts: They must have a high degree of accuracy and individual product specificity. In general terms, qualitative forecasting methods rely on managerial judgment; they do not use specific quantitative models. Qualitative methods are useful when there is a lack of data or when past data are not reliable predictors of the future. There are two general types of quantitative forecasting methods: time-series and causal forecasting. Quantitative methods utilize an analytical model to arrive at a forecast. The basic assumption for all quantitative forecasting methods is that past data and data patterns are reliable predictors of the future. Forecasting relies on predictive analytics, constructing a useful forecasting model from past demand and other relevant data.

In the remainder of this chapter, we refer to long, medium, and short time ranges. “Long range” will mean two years or more into the future, a common horizon for the planning of facilities and processes. “Medium range” is between six months and two years, the normal time frame for aggregate planning decisions, budgeting, and other resource acquisition and allocation decisions. “Short range” refers to less than six months, where the decisions involve procurement of materials and scheduling of particular jobs and activities.

10.2 QUALITATIVE FORECASTING METHODS Qualitative forecasting methods utilize managerial judgment, experience, and relevant data, if available. Because judgment is used, two different managers using qualitative methods may arrive at widely different forecasts. Some people think that qualitative forecasts should be used only as a last resort. This is not strictly true. Qualitative forecasts should be used when past data are not reliable indicators of future demand, for example, when changes in styles or technologies alter customer preferences. Qualitative forecasting must also be used for new-product and new-service introductions for which historical demand data are not available. In these cases, qualitative methods can be used to develop a forecast by life-cycle analogy or by the use of market research data. Note, a systematic approach to qualitative forecasting is possible even though judgment and experience is used.

10.3 TIME-SERIES FORECASTING Time-series methods make detailed analyses of past demand patterns to predict demand in the future. One of the basic assumptions of all time-series methods is that demand can be decomposed into components such as average level, trend, seasonality, cycle, and random error. A sample of these components for a representative time series is shown in Figure 10.1. When the components are added together (or in some cases multiplied), they will equal the original time series data pattern.

FIGURE 10.1 Decomposition of time-series data.

The basic strategy used in time-series forecasting is to identify the magnitude and form of each component on the basis of available past data. These components, except the random error component, are then used to estimate future demand in the form of a forecast. In discussions of time-series forecasting, the following symbols and terminology are used:

1. Dt = demand during period t 2. Ft+1 = forecast demand for period t + 1 3. et = Dt – Ft = forecast error in period t 4. At =average computed through period t We are at the end of period t, having just observed actual demand that defines the value of Dt, and are making forecasts for future periods t + 1, t + 2, t + 3, and so on. 10.4 MOVING AVERAGE The simplest analytics method of time-series forecasting is the moving average. For this method, it is assumed that the time series has only a level component plus a random error component. No seasonal pattern, trend, or cycle components are assumed to be present in the demand data. More advanced versions of the moving average can, however, include these additional components. When the moving average is used, a given number of periods (N) is selected for the computations. Then the average demand, At, for the past N periods at time t is computed as follows:

Since we are assuming that the time series is level (or horizontal), the best forecast for period t + 1 is simply the average demand observed through period t. Thus, we have

Each time Ft + 1 is computed, the most recent period of demand is included in the calculation of the average and the oldest period of demand is dropped. This procedure maintains N periods of demand in the forecast and lets the average move along as new demand data are observed. In Table 10.3, a three-period moving average is used for forecasting purposes. Notice how the moving average is offset by one period to obtain the moving forecast. The forecast error is also shown in the table as the difference between actual and forecast demand. Always use the forecast for period t (Ft) in computing forecast errors, not the average for period t (At).

10.5 EXPONENTIAL SMOOTHING The second predictive analytics method for forecasting is useful for reducing the amount of past demand data that must be carried forward. Exponential smoothing is based on the simple idea that a new average can be computed from an old average along with the most recent observed demand. Suppose, for example, we have an old average of 20 and we have just observed a demand of 24. The new average will lie between 20 and 24, depending on how much weight we want to assign to the newest demand versus the weight on the old average.

To answer the question of which is the best forecast, we need to look at forecast errors over many periods. Two measures of forecast accuracy are computed in Table 10.4 for 15 periods. One measure is simply the arithmetic sum of all errors, which reflects the bias in the forecasting method. Ideally, this sum should be zero, since the positive and negative errors should cancel out over time. In Table 10.4, both methods have a positive bias, with α = .1 producing more bias than α = .3. The second measure of forecast error is the absolute deviation. In this case the absolute values of the errors are summed, so that negative errors do not cancel positive errors. The result is a measure of variance in the forecasting method. The total absolute deviation for α = .1 is less than for α = .3.

10.6 FORECAST ACCURACY When exponential smoothing is used, an estimate of forecast accuracy should be computed along with the smoothed average. This accuracy estimate might be used for several purposes: 1. To monitor erratic demand observations or outliers, which should be carefully

evaluated and perhaps excluded from data analysis. 2. To determine when the forecasting method is no longer tracking actual demand and

needs to be reset. 3. To determine the parameter values (e.g., N and α) that provide the forecast with the

best accuracy. There are four ways to measure long-run forecast accuracy over several periods. (Recall that et = Dt – Ft is the forecast error for period t.)

Note that n is the number of past periods used to compute the cumulative error measurements. The second use of MADt is to determine whether the forecast is tracking with the actual time- series values. To determine this, a tracking signal is computed, as follows:

The tracking signal is a ratio of bias (cumulative forecast error) in the numerator divided by the most recent estimate of MADt. If demand variations are random, control limits of ±6 on the tracking signal should ensure only a 3 percent probability that the limits will be exceeded by chance.1 Thus, when the tracking signal exceeds ±6, the forecasting method should be reset to more nearly equal observed demand. In Table 10.4, the tracking signal does not exceed ±6 in any period. Therefore, the forecast is considered to be tracking sufficiently close to actual demand. As an example of these computations, refer to Table 10.4. In the last two columns of the table we have computed smoothed MADt and tracking signal. Starting with the arbitrary assumption that MAD0 = 7, we can compute MAD1 using α = .3:

10.7 ADVANCED TIME-SERIES FORECASTING A variation of exponential smoothing is adaptive exponential smoothing. In one form, simple exponential smoothing is used but the smoothing coefficient is varied at each forecast by ±.05 to determine which of the three forecasts has the lowest forecast error. The resulting value of α is used for the next-period forecast. Another type of adaptive smoothing is to continually adjust α on the basis of current forecast error. If there is a

large forecasting error, α will be large until the forecast comes back on track. When the error is smaller, α will also be small and a stable forecast will result. This method appears to work quite well for inventory forecasting situations. Table 10.5 summarizes four time-series forecasting methods. We have already discussed two of them, moving average and exponential smoothing. The remaining two are described briefly below. TABLE 10.5 Time-Series Forecasting Methods

Accuracy

Time-

Series

Methods

Description of

Method Uses

Short

Term

Medium

Term

Long

Term

Relative

Cost

1. Moving

averages

Forecast is

based

on arithmetic

average or

weighted

average of a

given number of

past data points.

Short- to

medium-

range planning

for inventories,

production

levels, and

scheduling.

Poor to

good Poor

Very

poor Low

2.

Exponential

smoothing

Similar to

moving average,

with

exponentially

more weight

placed on recent

data. Well

adapted to large

number of items

to be forecast.

Same as

moving

average.

Fair to

very

good

Poor to

good

Very

poor Low

3.

Analytical

models

A linear or

nonlinear model

fitted to time-

series data,

usually by

regression

methods.

Limited, due to

expense, to a

few products.

Very

good

Fair to

good

Very

poor Medium

4. Box-

Jenkins

Autocorrelation

methods are

used to identify

underlying time

series and to fit

the “best”

model. Requires

about 60 past

data points.

Limited, due to

expense, to

products

requiring very

accurate short-

range

forecasts.

Very

goodto

excellent

Fair to

good

Very

poor

Medium

to high

Source: Exhibit adapted from David M. Georgoff and Robert Murdick, “Manager’s Guide to Forecasting,” Harvard Business

Review, January–February 1986, pp. 110–120.

At Kellogg USA, demand for cereal is translated into forecasts, useful for inventory

planning and meeting supply chain goals.© Rick Souders/Getty Images

A customized analytical model can be fitted to a time series with level, trend, and seasonal components. For example, a linear regression model or nonlinear methods can be used. In some cases, the resulting model may provide a more accurate forecast than exponential smoothing. However, a custom-fitted model is more expensive because an individual with sophisticated skills will be used to develop it, and so the trade-off between accuracy and model cost must be considered. Another option, the sophisticated Box-Jenkins method, has been developed for time- series forecasting. This technique permits more precise analysis of proposed models than is possible with the other methods. The Box-Jenkins method requires about 60 periods of past data and is too expensive to use for routine forecasting. For a special forecast involving a costly decision, however, the use of Box-Jenkins may be warranted. In summary, time-series methods are useful for short- or medium-range forecasts when the demand pattern is expected to remain relatively stable. Time-series forecasts are often inputs to decisions concerning aggregate output planning, budgeting, resource allocation, inventory, and scheduling.

10.8 CAUSAL FORECASTING METHODS The second type of quantitative forecasting, causal forecasting methods, develop a cause- and-effect model between demand and other variables. For example, the demand for ice cream may be related to population, the average summer temperature, and time. Data can be collected on these variables and analysis conducted to determine the validity of the proposed model. One of the best-known causal methods is regression, which is usually taught in statistics courses. For regression methods, a model must be specified before the data are collected and the analysis is conducted. The simplest case is the following single-variable linear model:

where  yˆy^ = estimated demand  x = independent variable (hypothesized to cause yˆy^)  a = y intercept  b = slope Data are collected for x and y in this model, and the parameters a and b are estimated. Then estimates of demand can be made from the above equation. Of course, more complicated models with multiple x’s can be developed.

10.9 SELECTING A FORECASTING METHOD In this section, we present a set of factors to consider for selecting a forecasting method.

1. Use or decision characteristics. The forecasting method must be related to the use or decisions required. The use, in turn, is closely related to characteristics such as accuracy required, time horizon of the forecast, and number of items to be forecast. For example, inventory, scheduling, and pricing decisions by a big box retailer require highly accurate short-range forecasts for a large number of items. Time-series methods are ideally suited to these requirements. By contrast, decisions by an auto manufacturer involving process, facility planning, and marketing programs are long range in nature and require less accuracy. Qualitative or causal methods tend to be more appropriate for those decisions. In the middle time range are aggregate planning, capital-budgeting, and new- product and new-service introduction decisions, which often utilize time-series or causal methods. 2. Data availability. The choice of forecasting method is often constrained by available data. An econometric model may require data that are simply not available in the short run; therefore, another method must be selected. The Box-Jenkins time-series method requires about 60 data points (five years of monthly data). In some cases, data can be collected, but then time and resources must also be considered relative to the importance of the forecast. The quality of the data available is also a concern. Poor data lead to poor forecasts. Data should be checked for extraneous points or outliers. 3. Data pattern. The pattern in the data will affect the type of forecasting method selected. If the time series is level, as we have assumed in most of this chapter, a fairly simple method can be used. However, if the data show trends or seasonal patterns, more advanced methods will be needed. The pattern in the data will also determine whether a time-series method will suffice or whether causal models are needed. If the data pattern is unstable over time, a qualitative method may be selected. One way to detect the pattern is to plot the data on a graph as the first step in forecasting. An issue concerning the selection of forecasting methods is the difference between fit and prediction. When different models are tested, it is often thought that the model with the best fit to historical data (least error) is also the best predictive model. This is not true. For example, suppose demand observations are obtained for the last eight time periods and we want to fit the best time-series model to these data.

10.10 COLLABORATIVE PLANNING, FORECASTING, AND REPLENISHMENT Collaborative planning, forecasting, and replenishment (CPFR) is sharing information between business customers and suppliers in the supply chain during the planning and forecasting process. For example, a customer (e.g., retailer) may have information on planned sales promotions that are not known to the supplier. In this case a supplier’s forecast based on time- series data alone would be inaccurate, but it could be adjusted if the retailer information were made available.

في نهاية الوحدة )11( من النسخة القديمة يوجد تمارين محلولة ارجعوا لها لإلستزادة هي نفسها من كتاب

النسخة الجديدة

Discussion Questions 1. Is there a difference between forecasting demand and forecasting sales? Can demand

be forecast from historical sales data? 2. What is the distinction between forecasting and planning? 3. Qualitative forecasting methods should be used only as a last resort. Agree or

disagree? Comment. 4. Describe the uses of qualitative, time-series, and causal forecasts. 5. Qualitative forecasts and causal forecasts are not particularly useful as inputs to

inventory and scheduling decisions. Why is this statement true? 6. What type of time-series components would you expect for the following products

and services? a. Monthly sales of a retail florist. b. Monthly sales of milk in a supermarket. c. Daily demand in a call center.

7. What are the advantages of exponential smoothing over the moving average and the weighted moving average?

8. How should the choice of α be made for exponential smoothing? 9. Describe the difference between fit and prediction for forecasting models. 10. In the Stokely Company, marketing makes a sales forecast by developing a sales

force composite. Meanwhile, operations makes a forecast of sales based on past data, trends, and seasonal components. The operations forecast usually turns out to be 20 percent less than the forecast of the marketing department. How should forecasting in this company be done?

11. Explain how CPFR can be used to reduce forecasting error. 12. Under what circumstances might CPFR be useful, and when is it not useful?