Stevenson 6ce ISM Chapter 03 PDF

Preview

Summary

solutions to chapter 3 problems ...

Description

CHAPTER 3 FORECASTING Teaching Notes This is a fairly long chapter, so you may want to be selective about the topics covered. I tend to focus on time series models, including trend and seasonality. To solve numerical problems, the “black box” Excel spreadsheets on the book’s Website can be used. However, if the objective of the course includes improving the Excel skills of the students, the instructor could set up simple worksheets of examples (e.g., by copying the imbedded worksheets in the electronic version of this manual and pasting it in Excel) and provide these to students to change for their problems. Excel can also be used for drawing nice charts.

Answers to Discussion and Review Questions 1.

Quantitative techniques lend themselves to computerization, they are more objective, and they may force the forecaster to collect more information. All these result in more accuracy vs. judgmental methods. On the other hand, in many cases, insufficient data forces one to use judgmental methods which can also include direct demand information such as the intention of a customer to buy more or our intention to raise the item’s price.

2.

(a) Influencing factors in the past continue to influence the demand in the future (this is an assumption), (b) Forecasts are intrinsically inaccurate; need to also estimate a measure of inaccuracy such as MAD, (c) Forecasts for a group of items will be more accurate, because the highs and lows cancel each other, as in central limit theorem (d) Forecast accuracy decreases the farther the forecast is in the future, because the factors will have more time to change.

3.

(a) The forecasting horizon must be long enough so that its results can be used. (b) The degree of accuracy of the forecast should be stated, because then one will know how much one can rely on it. (c) The forecasting method/software chosen should be reliable; it should work consistently. (d) The forecast should be expressed in meaningful units. Financial planners need to know demand in dollars, whereas demand and production planners need to know demand in units. (e) All functions of an organization should be using the same forecast, because then it will be possible to meet it. (f) The forecasting technique should be simple to understand and use, because the management will want to understand it, and the forecaster will be able to use it correctly and quickly.

4.

Long term; e.g., whether or not to introduce a new product Medium term; e.g., aggregate planning Short term; e.g., staff requirement and scheduling

5.

(a) Determine the purpose of the forecast, the level of detail required, the amount of resources (personnel, computer time, dollars) that can be justified, and the level of accuracy necessary. (b) Establish a forecasting horizon. (c) Gather and analyze relevant historical data. Ensure that the data is of past demand rather than sales or shipments, which will be different if there were stock-outs. Identify any assumptions that are made. (d Select a forecasting technique. (e) Prepare the forecast. (f) Monitor the forecast. If it is not performing in a satisfactory manner, re-examine the parameters of the technique or use a different technique.

3-1 Instructor’s Manual, Chapter 3

6.

Shipments indicate how much the customer bought, while demand indicates how much they wanted. The distinction is important when demand exceeds shipments. In this case, using the shipments or sales data will underestimate the demand.

7.

Flexible production systems—those that can respond quickly to changes in demand—require a shorter forecasting horizon and, hence, benefit from more accurate short-term forecasts than competitors that are less flexible and who must therefore use longer forecasting horizons.

8.

Poor forecasting leads to poor planning. This could result in offering products and services that customers do not want. Poor forecasting and planning would negatively affect budgeting and planning for capacity, sales, production and inventory, labor, purchasing, energy requirements, capital requirements, and materials requirements.

9.

Sales indicate how much customers bought, while demand indicates how much they wanted. The distinction is important when demand exceeds supply, because supply places an upper bound on the data

10.

Executive Opinions: A small group of upper-level managers (e.g., VPs of marketing, operations, and finance) may meet and collectively develop a forecast. Sales Force Opinions: The sales staff or the customer service staff is often aware of any plans that the customers may be considering for the future, including the current level of customer inventory. Consumer Surveys: Potential consumers’ opinion are sought either using a questionnaire (by mail or phone) or through group meetings (focus groups). Historical Analogies: The demand for a similar product in the past, after some adjustment, can be used to forecast a new product’s demand. Expert Opinions (Delphi method). It involves circulating a series of questionnaires among experts. Responses are kept anonymous, which tends to encourage honest responses and reduces the risk that one person’s opinion will prevail. Each new questionnaire is developed using the information extracted from the previous one, thus enlarging the scope of information on which participants can base their judgments. The goal is to achieve a consensus forecast.

11.

A time series is a time-ordered sequence of observations taken at regular intervals of time.

12.

Time series models forecast on the assumption that future values of the series can be estimated from its own past values. No attempt is made to identify variables that influence the series.

13.

The Naïve method will use the sales during the same hour of the same day last week.

14.

Moving average averages a number of recent actual values as forecast for current period. It is updated as new values become available. Exponential smoothing uses the previous forecast plus a percentage of the difference between that forecast and the previous actual value to forecast the current period. The averaging methods are not suitable for data with seasonality and/or trend because they will always lag behind these patterns.

15.

Exponential smoothing (a) requires less data storage and is easier to use (the updating formula), (b) gives more weight to recent data, and (c) is easier to fit to data (i.e., determining the right α is easier than determining the no. of periods in the moving average).

16.

The fewer the periods in a moving average, the greater the responsiveness.

17.

The larger the smoothing constant in exponential smoothing, the greater the responsiveness.

3-2

Operations Management, 6/C/e

18.

The period number (t) is used as the independent (x), i.e., y = a + bt.

19.

The exponential trend equation is

bt

y=ae

, where a is the intercept (= the value f y when t = 0), and

b

100×( e −1) is the percentage growth rate. 20.

Trend-adjusted forecast for period t+1:

TAF t+1 = St +T t

Smoothed series at the end of period t:

S t =TAFt +α ( A t −TAF t )

T t =T t−1 +β ( St −S t−1 −T t−1 ) Smoothed trend at the end of period t: In the trend-adjusted exponential smoothing, the series is smoothed using simple exponential smoothing,

S −S

as in the case of no trend. However, the current trend in the smoothed series ( t t−1 ) is also computed and smoothed using simple exponential smoothing. Finally, the sum of the smooth series and smooth trend is used to forecast the next period. 21.

A seasonal relative or index of a season is the proportion of the season’s demand relative to the average season’s demand in the year. For example, retail sales in the 4th quarter may have 1.20 as seasonal relative, meaning that sales during the 4th quarter is on average 20% higher that an average season’s sales.

22.

Centred moving average (CMA) is a moving average positioned at the centre of the data that were used to compute it. Each centred moving average is based on all the different seasons of the year around it (essentially 4 quarters) and therefore includes the highs and lows associated with all seasons. Thus, seasonality is eliminated in the CMA.

23.

The ratio of each season’s demand to its CMA is computed (if the CMA exists). Then, the ratios of same seasons (from different years, e.g.,) are averaged. Finally, these averages are rescaled so that they add up to the total number of seasons.

24.

(a) (b) (c) (d) (e)

25.

Causal (regression) model uses an equation to forecast the demand (y) of an item: e.g., y = a + bx, where x is the explanatory variable. After a and b are estimated using past data, x has to be forecasted for the next period, then substituted in the equation to get the forecast of y for the next period. It is essential that forecasting x is easier (e.g., when x is under our control as in future price an item).

26.

Correlation coefficient is a measure of the strength of relationship between two variables. If a causal model is to be used to forecast the demand (y) of an item, then it is important that the relationship between x and y variables is strong. Negative correlation coefficient means that x and y are inversely related. Variable x can still be used to forecast y, as long as the correlation coefficient is close to -1.

27.

Forecasting accuracy is the degree of correctness of the forecasts generated by the forecasting process. It is easier to measure inaccuracy, i.e., the degree of incorrectness of the forecasting process. The incorrectness of a forecast is measured by the forecast error (actual – forecast). For a forecasting process, the forecast errors for a long-enough period are used, in an aggregate form, as its measure of (in) accuracy. Three common aggregate measures of (in) accuracy are MAD, MSE, and MAPE (defined in the next question).

Compute the seasonal relatives (SRs). Deseasonalize the demand data (divide demand by SRs). Fit a trend model to the deseasonalized demand data. Forecast using this model (to obtain the deseasonalized forecasts). Reseasonalize the deseasonalized forecasts (multiply by SRs).

3-3 Instructor’s Manual, Chapter 3

28.

MAD is the average of absolute value of forecast errors, MSE is the average of square forecast errors, and MAPE is the average of absolute value of percentage of error (relative to demand). MSE gives considerably more weight to large forecast errors than MAD or MAPE. MAPE is scale independent, making it easier to understand.

29.

Control chart for forecast errors is a time series plot of forecast errors that has limits for individual forecast errors. Control limits reveal the bounds of random forecast errors (as a result of random demand variations); they enable managers to judge if a forecasting technique is not performing well (and hence, when a technique should be re-evaluated).

30.

The relative costs of re-evaluating a forecasting technique when nothing is wrong (type I error) versus not re-evaluating it when something is wrong (Type II error).

31.

Tracking signal (TS) is a measure used to control the forecasting process; TS = sum of forecast errors divided by mean absolute forecast error. TS uses the ratio of sum of forecast errors over MAD whereas control chart uses the individual forecast errors. Thus, TS uses bias, the cumulative forecast error, whereas control chart uses each forecast error.

32.

TS directly monitors the bias because TS = bias / MAD. To monitor bias using control chart for forecast errors, one needs to pay attention to the line plot of forecast errors. If the line plot is consistently over or under the zero line, then bias is increasing in absolute value.

33.

34.

A reactive approach takes the forecast as a “given” while a proactive approach takes an unacceptable forecast and attempts to alter the demand. An example of the reactive approach is preparing for a hurricane. An example of the proactive approach is a college or university adopting a more aggressive stance towards enrolment due to a forecast of declining number of applicant. Generally, organizations that use advertising, promotions, discounts, and so on, tend to be proactive in dealing with forecasts. a.

naïve (last year’s demand on Mother’s Day) or averaging demand during past few Mother’s Days (moving average). b. consumer surveys c. consumer surveys d. averaging (moving average, exponential smoothing; assuming no trend)

35.

a. Consumer surveys may be invalid if they are not carefully constructed, administered, and interpreted. Moreover, respondents may be ill informed or otherwise formulate answers that do not correctly reflect their future actions. b. Salespeople often tend to be overly optimistic or pessimistic. They may attempt to use estimates to influence quotas. c. Committees of managers or executives can be expensive, diffuse responsibility for a forecast, and reflect the opinions of a few dominant members.

Answers to Taking Stock 1.

3-4

If the forecast system is too responsive and it becomes too sensitive to the changes in actual demand, it will have a tendency to overreact, resulting in too much adjustment to the forecasted demand. On the other hand, if the forecasts are too stable, they do Operations Management, 6/C/e

not react fast enough to changes in demand, resulting in insufficient response to changes in actual demand.. 2.

Forecasting needs to be a collaborative effort involving the forecaster, marketing/sales, finance, and production people.

3.

More sophisticated software and faster computers have enabled more sophisticated forecasting techniques (such as Box Jenkins method). Also, EDI and Internet lines have enabled collecting data and collaborative forecasting.

4.

(a)

Establish a forecasting process, guidelines and rules

(b)

Require the assumptions and data used and reasons

(c)

Compare the forecast with forecasts from other sources

(d)

Use quantitative techniques; look at the data over a longer period

(e)

Try to indentify the bias (e.g., using control chart or tracking signal)

(f)

Try to identify the source of bias

(g)

Look at signals, e.g., withholding info, requests to change the forecast

(h)

Set up a forecast file to keep a paper trail

(i)

Develop ethical code of conduct and a clear statement of consequences of forecast bias

(j)

Conduct meetings and workshops to make the code of conduct part of org. culture

Answers to Critical Thinking Exercise 1. Although understandable, Omar’s approach is not ethical. He should turn in the forecast based on the information he has and tell his superiors that he thinks he can get those numbers up. The only pro to Omar’s optimistic forecast might be preventing Omar from being laid off over the near-term if he can convince customers not to cut back on orders. The cons are that if sales do not materialize, Omar will be laid off and inventories are going to be high at both his company and at his customers. 2. The conditions that would have to exist for driving a car that are analogous to the assumptions made when using exponential smoothing are that the immediate future will be like the recent past. This would suggest no sharp curves or turns on the road; constant traffic conditions; no traffic lights or stop signs; constant road conditions. 3. Potential investors would expect information on the current and future size of the market, the expected initial market share and growth rate for 5-10 years, profit/loss projections for the forecast horizon, and the likelihood of achieving the projected results. 4. Generally, the more time is spent collecting data and information, and the more sophisticated the forecasting technique used, the more accurate the forecast will be. To determine the right balance, one needs to estimate the cost of consequence of forecast error, and the cost of improving the accuracy.

Answers to Experiential Learning Exercise 1. 2.

From 2012 to 2017, there is a positive trend, but there is no obvious seasonality in the graph. This exercise requires collecting the Short Term Temperature forecasts for the same and next day during 4 afternoons in a row (e.g., see the middle of http://www.theweathernetwork.com/weather/cask0276). For example, Sun: collect Mon Forecast; Mon: collect Mon Actual, Tue Forecast; Tue: collect Tue Actual, Wed Forecast; Wed: collect Wed

3-5 Instructor’s Manual, Chapter 3

Actual. Then, the actual, next-day forecast, and naïve forecast for each day, Mon to Wed are set up like the data in Solved Problem 8 and solved the same way. 3.

This exercise requires collecting the Short Term Temperature forecasts for the same and next day (e.g., see the middle of http://www.theweathernetwork.com/weather/cask0276) and Long Term Temperature, High, forecasts for two days and 3 days ahead during 6 afternoons in a row (e.g., see the bottom of http://www.theweathernetwork.com/weather/cask0276). For example, Sun: collect Mon Forecast, Tue Forecast, Wed Forecast; Mon: collect Mon Actual, Tue Forecast, Wed Forecast, Thur Forecast; Tue: collect Tue Actual, Wed Forecast, Thur Forecast, Fri Forecast; Wed: collect Wed Actual, Thur Forecast, Fri Forecast, Sat Forecast; Thur: collect Thur Actual; Fri: collect Fri Actual; Sat: collect Sat Actual. For each set of 1-day, 2-day, and 3-day forecasts, the problem should be set up like Example 12 and a measure such as MAD calculated. Note that each problem will have different periods: For example, 1-day forecasts will be Mon-Thur, 2-day forecasts will be Tue-Fri, and 3-day forecasts will be Wed-Sat.

4.

This problem is similar to Ex. 2 above, but instead of forecasts from the Weather Network, students are supposed to pick a forecasting method and perform next-day forecasts themselves. Given the short-term nature of this exercise (next-day forecasts for 3 days), the best forecasting methods are averaging methods such as the moving average.

Answer to Internet Exercise From 2011 to 2017, there seems to be a negative relationship between crude oil prices and USD/CAD. This means an increase in crude oil prices will increase the value of CAD. The relationship seems strong, so oil prices are a good predictor of CAD.

Solutions to Problems 1. a&b.

Plotting each data series (see below) reveals that blueberry muffin sales are stable, varying around an average (constant). Therefore, the naive forecast for workday 16 should be the last value, 33 dozens. The demand for cinnamon buns has an increasing trend. The last change was from 31 to 33 (33-31 = 2). Using the last value and adding the last trend change, the forecast for workday 16 is 33 + 2 = 35 dozens. Demand for cupcakes has seasonal variation with peaks every five workdays. Workday 1 = 45, Workday 6 = 48, Workday 11 = 47 are peak workdays. There is no trend. Using Workday 11’s sales, the forecast for Workday 16 is 47 dozens.

60 50 40 Blueberry Muffins

30

Cinnamon buns Cupcakes

20 10 0 1

3-6

2

3