MAB_J8

LDR 5301, Methods of Analysis for Business Operations 1

Course Learning Outcomes for Unit VIII Upon completion of this unit, students should be able to:

7. Assess the differences between correlation and causation. 7.1 Justify the selection of a forecasting model over other techniques in a given scenario. 7.2 Develop a forecasting model.

Course/Unit

Learning Outcomes Learning Activity

7.1

Unit Lesson Chapter 5 Video Segment: Correlation and Causation: Illustrating the Difference Video Segment: Correlation and Causation Video Segment: Forecasting in Business Unit VIII Assignment

7.2

Unit Lesson Chapter 5 Video Segment: Forecasting in Business Unit VIII Assignment

Required Unit Resources Chapter 5: Forecasting In order to access the following resources, click the links below. ClickView Pty Limited (Producer). (2013). Correlation and causation: Illustrating the difference (Segment 6 of

7) [Video]. In. Epidemiology: Linking smoking and lung cancer. Films on Demand. https://libraryresources.columbiasouthern.edu/login?auth=CAS&url=https://fod.infobase.com/PortalPl aylists.aspx?wID=273866&xtid=56488&loid=245471

Palomar Community College District (Producer). (2010). Correlation and causation (Segment 3 of 6) [Video].

In Correlation and regression: Lecture 10. Films on Demand. https://libraryresources.columbiasouthern.edu/login?auth=CAS&url=https://fod.infobase.com/PortalPl aylists.aspx?wID=273866&xtid=150110&loid=506159

TV Choice Ltd. (Producer). (2011). Forecasting in business (Segment 5 of 9) [Video]. In Accounting & finance

clips 1: Accounting, forecasting, and breakeve. Films on Demand. https://libraryresources.columbiasouthern.edu/login?auth=CAS&url=https://fod.infobase.com/PortalPl aylists.aspx?wID=273866&xtid=128753&loid=450483

The transcripts for these videos can be found by clicking on “Transcript” in the gray bar to the right of the video in the Films on Demand database. Unit Lesson

Introduction to Forecasting Welcome to Unit VIII! You made it to the end of the course. In this unit, we will examine forecasting and causation techniques. What is the intent of a forecast? The intent is to predict with reasonable accuracy

UNIT VIII STUDY GUIDE Causation and Forecasting

LDR 5301, Methods of Analysis for Business Operations 2

UNIT x STUDY GUIDE Title

(ideally, 100%) that your decision-making process for future action is good. According to Stevenson (2018), elements of a good forecast should be

• timely, • reliable, • accurate, • meaningful, • written, and • easy to use.

Most of us are familiar with the basic forecast we see on The Weather Channel regarding what type of weather we will have in our area; it helps us effectively and efficiently plan our day. If we do not watch The Weather Channel, most of us will pull out our smartphones and click on the weather for our city or town, which will provide information for the area. However, we all know the weather is hard to predict; it is a moving event. Imagine that you plan to work in the yard with a forecast that is sunny, and then suddenly the clouds come in and release a 30-minute downpour of rain. What we may not know about forecasting are the components and processes that go into making future predictions. At times, it may seem like forecasters are using a crystal ball, but actually, they are using very complex algorithms to determine future trends based on past trends, current data, previous data, sentiment, and consumer behavior analysis.

Forecasting Methods There are forecasting methodologies and processes that aid the forecaster. The big question with forecasting is: “What do I do?” To help forecasters with this conundrum, there are models and tools to assist (see Figure 1).

Forecasting Techniques (Render, et al., 2018)

LDR 5301, Methods of Analysis for Business Operations 3

UNIT x STUDY GUIDE Title

Let’s begin by outlining the major types of models and how they are used. Forecasters have a choice, and depending on what is desired, whichever methods that best fits the needs of the forecaster can be chosen.

Qualitative Methods Qualitative methods are made up of four different forecasting techniques: Delphi method, jury of executive opinion, sales force composite, and consumer market survey. You can review each of them in detail in the textbook. The three that apply to most of us in our work environments are the jury of executive opinion, sales force composite, and the consumer market survey. This is because they deal with everyday decisions based on product development, manufacturing processes, and consumer demand. Think about automobile manufacturing for a moment. If you were in the jury of executive opinion, what would you want to know to forecast the correct number of cars to produce? You would look at regional data based on employment, income, consumer-spending habits, and previous automobile sales by region. All of this goes into a decision-making model that we reviewed in a previous unit. The sales force composite contributes to the jury of executive opinion because these are the regional sales individuals of car dealerships who know their product, what is selling, why it is selling, and how many sales they anticipate in the next production cycle. Finally, the consumer market survey (data) is provided to the regional sales managers and data crunched by the jury of executive opinion. The consumer data is basically surveys about the product, what the consumers like, what they did not like, and/or what they would like to see in future products. This helps managers and engineers develop the next product. In this example of automobile manufacturing, you can see how each technique feeds into the other to provide a more accurate forecast.

Time Series Methods

These methods are based on a record of data that is graphed over a time period. Review Figure 2. You can see the basic chart has sales on the vertical axis and the time period on the horizontal access. You should be familiar with these charts from the regression model unit where distributions were placed on vertical and horizontal axis. The purpose here is to place the data points and determine a trend (correlation) or randomness in the data.

What every forecaster wants is a very tight straight trend line as seen in the Series 3 middle trend line. Why? It is easier to predict the next period of sales with little deviation versus the Series 2 trend line. As you can see in Figure 2, the word seasonal has appeared. Now we know there are seasons, and with each season forecasters can determine how many bikinis to make, or how many winter coats to make. Make too many, and retailers are stuck with excess product on the shelf. Manufacturers might make too few coats because the winter forecast predicted warm weather, but instead a two-month blizzard occurred, and production did not anticipate the need for sweaters, mittens, hats, and insulated coats.

Figure 2 (Render et al., 2018)

LDR 5301, Methods of Analysis for Business Operations 4

UNIT x STUDY GUIDE Title

Causal Methods Finally, the causal methods use regression models that we have looked at previously in Chapter 4 (models that have perfect positive correlation, positive correlation, no correlation, and perfect negative correlation). In Figure 3, you can see how quantitative analysis depends on data, and how to use the proper tools to analyze the data to make a decision. Recalling information from Unit VII, we can see below, the correlation coefficient at work with the variables plotted and the r values. Note that in Figure 3 (a), (b), and (d) displays, an almost perfect correlation in some instances. The benefit of this is that both variables are dependent on each other. This provides the closeness to the regression line.

Measures of Forecast Accuracy

As a future forecaster, you want to ensure that the forecast you have provided is accurate. You would not want to produce a forecast that is only 25% reliable, which means that 75% of the time you were incorrect. If this were your forecasting track record, you just might be out of a job, or, as you will see in this unit’s assignment, make a poor recommendation on a stock investment. To assist forecasters, measures of forecast accuracy were developed; these include the mean absolute deviation (MAD). Render et al., (2018) noted that the MAD “is computed by taking the sum of the absolute values of the individual forecast errors and dividing by the number of errors” (p.151). Refer to Table 5.1 in the textbook to see how the MAD is computed. After all the number crunching in that table occurs, the MAD is 17.8, which means that the proposed forecast missed the actual value of units needed by 18 (rounded up). The impact of that is cost and inventory on hand.

To assist in computing forecast accuracy in a time series, forecasting models are used (Render et al., 2018). The most popular analysis tools used in a forecasting model are moving averages. What the moving average does is smooth out the data, which takes out the large spikes and declines in the represented data so that a smooth line is represented to the current trend.

Moving Averages As mentioned above, moving averages assist the forecaster because they smooth out the data over a specific period. What develops is a trend line that trails the actual data. A great example of this is looking at the Dow Jones Industrial Average (DJIA) with a 200– and a 50–day moving average. Figure 4 displays a year-to-date daily chart from 2018 to 2019 of the DJIA closing index price. Using the moving average forecast formulas (Formulas 5-4 and 5-5) on page 154 of the textbook, the forecasters have developed a long-term trend (200

Figure 3 (Render et al., 2018)

LDR 5301, Methods of Analysis for Business Operations 5

UNIT x STUDY GUIDE Title

day) and a short-term trend (50 day) to determine price actions, and help their decision-making to either buy, hold, or sell stocks that are in the index. As you look at the figure what do you notice? You should notice:

• the erratic behavior of the index with mountains and valleys in the data; • how the moving average smooths out the data to develop a trend line; and • how the 200–day and 50–day averages provide information on trends. The 50–day, since shorter, will

always signal an earlier sell or buy signal. Finally, at the right side of the figure where the index has plunged downward, it is going to take time for the 50–day and 200–day averages to adjust to a new trend (either flat, upward-recovery, or continued downward). Notice that you can see the selloff in the index in late 2019 from a 26,000 level to a 22,859 level. This is clearly, a downward trend in actual prices, with the moving averages beginning to flatten and turn downward. Now look at Figure 5. The DJIA index has recovered, almost in a perfect V formation, but the 200– and 50–day moving averages lag behind the index. Why? Because it is a moving average, and as a new data point is added, an old data point is dropped to compute the 200– and 50–day averages. Hence, the moving average will always be a little behind.

Figure 4 (Adapted from Chen, 2018)

Figure 5 (Adapted from DeCambre, 2019)

LDR 5301, Methods of Analysis for Business Operations 6

UNIT x STUDY GUIDE Title

Market-savvy investors and big institutional investors (e.g., BlackRock Capital, Fidelity, T Rowe Price) employ the jury of executive opinion, meaning that they have offices that strictly analyze economic data, technical quantitative data, and fundamental company data (like earnings). All this data is provided to them in professional briefings with a lot of PowerPoint slides. These managers then look at large amounts of data such as stock prices, the economy, monetary policy, interest rates, volume of trades, stocks advancing, stocks declining, and the volatility index, as well as previous historical price actions, to make their next decision. Many managers likely do not know the exact bottom of a trend, but there are some savvy managers that are willing to take on risk given the low indicator in the chart. The takeaway from all of this is not to confuse you, but to illustrate an example that displays forecasting, data selection, data analysis and interpretation, and a moving average used to make decisions on the next direction of the index.

Weighted Moving Averages and Exponential Smoothing Weighted moving averages and exponential smoothing are in the same format as a normal moving average computation with exceptions in the design of the formulas. The formula for weighted moving average can be found on page 154 in the textbook, and the formula for exponential smoothing can be found on page 156. So, which is better? There is no right answer. Some would prefer exponential smoothing because it is a little more accurate, although it is a bit more intense to compute (by hand). However, the overall result and expectation is to smooth out the data noise and uncover an underlying trend so that decision-making is more effective.

Conclusion Forecasting is not easy! Your forecast is only as good as the data you put into your computation. It could be a guessing game, or it can be close to the mark. Forecasting requires a good building block method to support the models. Elements of a good forecast should be timely, reliable, accurate, meaningful, written, and easy to use. In this unit, we have outlined and explained the different types of forecasting techniques and the models associated with each. Within each model, there were different techniques used for executive decisions by managers based on data, sales manager predictions based on previous sales and future demand, and consumer behavior analysis. In the end, your goal should be to provide a relatively accurate picture of where the trend is going with regard to the particular product, service, and stock price to improve the company or individual standing.

References Chen, J. (2018, December 20). Dow: Latest index to form a death cross. Investopedia.

https://www.investopedia.com/dow-latest-index-to-form-a-death-cross-4582105 DeCambre, M. (2019, March 8). The Dow is on the verge of a bullish golden cross, but stock-market analysts

aren’t exactly cheering. MarketWatch. https://www.marketwatch.com/story/the-dow-is-the-verge-of-a- bullish-golden-cross-but-stock-market-analysts-arent-exactly-cheering-2019-03-07

Render, B., Stair, R. M., Jr., Hanna, M. E., & Hale, T. S. (2018). Quantitative analysis for management (13th

ed.). Pearson. https://online.vitalsource.com/#/books/9780134518558 Stevenson, W. J. (2018). Operations management (13th ed.). McGraw-Hill.

LDR 5301, Methods of Analysis for Business Operations 7

UNIT x STUDY GUIDE Title

Suggested Unit Resources In order to access the following resources, click the links below. The Chapter 5 PowerPoint Presentation will summarize and reinforce the information from this chapter in your textbook. You can also view a PDF of the Chapter 5 presentation. Learning Activities (Nongraded) Nongraded Learning Activities are provided to aid students in their course of study. You do not have to submit them. If you have questions, contact your instructor for further guidance and information. For an overview of the chapter equations, read the Key Equations on pages 175–76 of the textbook. Then, complete the Self-Test on page 177–78 of your textbook to test your knowledge of concepts and terms in this unit. Use the key in the back of the book in Appendix H to check your answers. Finally, complete problem 5–40 on page 181 of the textbook to practice using the formulas you will need to know for the assignment in this unit. Use the answer key in Appendix G in the back of the textbook in order to check your answers