web viewteradata university and other hands-on exercises. team assignments and role-playing...

Chapter 4:Modeling and Analysis

Learning Objectives for Chapter 4

1. Understand the basic concepts of management support system (MSS) modeling2. Describe how MSS models interact with data and the user3. Understand some different, well-known model classes4. Understand how to structure decision making with a few alternatives5. Describe how spreadsheets can be used for MSS modeling and solution6. Explain the basic concepts of optimization, simulation, and heuristics, and when

to use them7. Describe how to structure a linear programming model8. Understand how search methods are used to solve MSS models9. Explain the differences among algorithms, blind search, and heuristics10. Describe how to handle multiple goals11. Explain what is meant by sensitivity analysis, what-if analysis, and goal seeking12. Describe the key issues of model management

CHAPTER OVERVIEW

In this chapter, we describe the model base and its management, one of the major components of decision support systems (DSS). We present this material with a note of caution: Modeling can be a very difficult topic and is as much an art as a science. The purpose of this chapter is not necessarily for you to master the topics of modeling and analysis. Rather, the material is geared toward gaining familiarity with the important concepts as they relate to DSS and their use in decision making. It is important to recognize that the modeling we discuss here is only cursorily related to the concepts of data modeling. You should not confuse the two. We walk through some basic concepts and definitions of modeling before introducing the influence diagram (see Online File W4.1), which can aid a decision maker in sketching a model of a situation and even solving it. We next introduce the idea of modeling directly in spreadsheets. We then discuss the structure and application of some successful time-proven models and methodologies: optimization, decision analysis, decision trees, analytic hierarchy process, search methods, heuristic programming, and simulation.

Copyright © 2011 Pearson Education, Inc. Publishing as Prentice Hall

2 Decision Support and Business Intelligence Systems (9th Edition) Instructor’s Manual

CHAPTER OUTLINE

4.1 OPENING VIGNETTE: MODEL-BASED AUCTIONS SERVE MORE LUNCHES IN CHILEA. PROBLEMB. SOLUTIONC. RESULTS

Questions for the Opening VignetteE. WHAT WE CAN LEARN FROM THIS VIGNETTE

4.2. MANAGEMENT SUPPORT SYSTEM MODELINGA. LESSONS FROM MODELING AT PROCTER & GAMBLEB. LESSONS FROM ADDITIONAL MODELING APPLICATIONS

Application Case 4.1: Lockheed Martin Space Systems Company Optimizes Infrastructure Project-Portfolio Selection

C. CURRENT MODELING ISSUES1. Identification of the Problem and Environmental Analysis2. Variable Identification3. Forecasting (Predictive Analytics)

Application Case 4.2: Forecasting/Predictive Analytics Boosts Proves to be a Good Gamble for Harrah’s Cherokee Casino and Hotel

4. Multiple Models5. Model Categories6. Model Management7. Knowledge-Based Modeling8. Current Trends in Modeling

Section 4.2 Review Questions

4.3. STRUCTURE OF MATHEMATICAL MODELS FOR DECISION SUPPORT

A. THE COMPONENTS OF DECISION SUPPORT MATHEMATICAL MODELS

1. Result (Outcome) Variables2. Decision Variables3. Uncontrollable Variables or Parameters4. Intermediate Result Variables

B. THE STRUCTURE OF MSS MATHEMATICAL MODELS Section 4.3 Review Questions

4.4 CERTAINTY, UNCERTAINTY AND RISKA. DECISION MAKING UNDER CERTAINTYB. DECISION MAKING UNDER UNCERTAINTYC. DECISION MAKING UNDER RISK (RISK ANALYSIS)


Decision Support and Business Intelligence Systems (9th Edition) Instructor’s Manual 3


4.5. MANAGEMENT SUPPORT SYSTEM MODELING WITH SPREADSHEETS

Application Case 4.3: Showcase Scheduling at Fred Astaire East Side Dance Studio Section 4.5 Review Questions

4.6. MATHEMATICAL PROGRAMMING OPTIMIZATION Application Case 4.4: Spreadsheet Model Helps Assign Medical Residents

A. MATHEMATICAL PROGRAMMINGB. LINEAR PROGRAMMINGC. THE LP PRODUCT-MIX MODEL FORMULATIOND. MODELING IN LP: AN EXAMPLE

Technology Insights 4.5: Linear Programming Section 4.6 Review Questions

4.7. MULTIPLE GOALS, SENSITIVITY ANALYSIS, WHAT-IF ANALYSIS, & GOAL SEEKINGA. MULTIPLE GOALSB. SENSITIVITY ANALYSIS

1. Automatic Sensitivity Analysis2. Trial-and-Error Sensitivity Analysis

C. WHAT-IF ANALYSISD. GOAL SEEKING

1. Computing a Break-Even Point by Using Goal Seeking Section 4.7 Review Questions

4.8. DECISION ANALYSIS WITH DECISION TABLES AND DECISION TREES

A. DECISION TABLES1. Treating Uncertainty2. Treating Risk

B. DECISION TREES Application Case 4.6: Decision Analysis Assists Doctor in Weighing Treatment Options for Cancer Suspects and Patients Section 4.8 Review Questions

4.9. MULTICRITERIA DECISION MAKING WITH PAIRWISE COMPARISIONS

A. THE ANSLYTIC HIERARCHY PROCESS Application Case 4.7: Multicriteria Decision Support for European

Radiation Emergency Support System Section 4.9 Review Questions



4.10. PROBLEM-SOLVING SEARCH METHODSA. ANALYTICAL TECHNIQUESB. ALGORITHMSC. BLIND SEARCHINGD. HEURISTIC SEARCHING

Application Case 4.8: Heuristic-Based DSS Moves Milk in New Zealand


4.11. SIMULATION Application Case 4.9: Improving Maintenance Decision

Making in the Finnish Air Force Through SimulationA. MAJOR CHARACTERISTICS OF SIMULATIONB. ADVANTAGES OF SIMULATIONC. DISADVANTAGES OF SIMULATIOND. THE METHODOLOGY OF SIMULATIONE. SIMULATION TYPES

1. Probabilistic Simulation2. Time-Dependent Versus Time-Independent Simulation3. Object-Oriented Simulation4. Visual Simulation

F. SIMULATION SOFTWAREG. SIMULATION EXAMPLES

Application Case 4.10: Simulation Applications Section 4.11 Review Questions

4.12. VISUAL INTERACTIVE SIMULATIONA. CONVENTIONAL SIMULATION INADEQUACIESB. VISUAL INTERACTIVE SIMULATIONC. VISUAL INTERACTIVE MODELS AND DSS


4.13. QUANTITATIVE SOFTWARE PACKAGES AND MODEL BASE MANAGEMENTA. MODEL BASE MANAGEMENT


4.14. RESOURCES, LINKS AND THE TERADATA UNIVERSITY NETWORK CONNECTIONA. RESOURCES AND LINKSB. CASESC. MILLER’S MIS CASESD. VENDORS, PRODUCTS, AND DEMOSE. PERIODICALSF. THE TERADATA UNIVERSITY NETWORK (TUN) CONNECTION



Chapter HighlightsKey TermsQUESTIONS FOR DISCUSSIONExercises

Teradata University and Other Hands-On ExercisesTeam Assignments and Role-Playing ProjectsInternet Exercises

End of Chapter Application Case: HP Applies Management Science Modeling to Optimize its Supply Chain and Wins a Major Award

Questions for the CaseReferences

TEACHING TIPS/ADDITIONAL INFORMATION The purpose of this chapter is to teach the concepts of modeling so students will begin to understand how models can be used with real world problems. It does so by going through a wide variety of model types, from the familiar (most DSS students will have encountered at least one financial model on a spreadsheet) through some that may have been seen in other courses (linear programming is often taught in operations management courses) to many that students will not have seen before. This can be a daunting amount of information to cover in the time you can allocate to a single chapter.

For this reason, you may want to select parts of this chapter to go into more deeply than others. Your choice of topics to stress will depend on your own perspective as to what is important, what isn’t. It’s still a good idea to at least touch on all the sections, though. Each one discusses a method that has been used in some DSS and that students may therefore encounter on the job. Your professional experience may tell you that some are less important than others, but that doesn’t mean any are totally unimportant.

Modeling is an applied discipline, best learned by reinforcement, through exercises and problems. Encourage the students to understand it through trial-and-error. The chapter mentions several software tools. You may wish to download, install, and use some of them. You can also work through a simple simulation by hand—for example, modeling waiting times and customers turned away at a barber shop with two barbers, two more chairs in the waiting area, and any distribution of arrival and service times you choose. Such a simulation can easily be done using a standard six-sided die to generate random numbers or by any of several other “low-tech” methods. If you have access to and are familiar with a simulation package, showing students how this simulation is programmed for that package will bring home the lessons of what simulation is really about.

You may also wish to read at least some of the articles cited in the text as examples. The text, for reasons of space, provides only a brief summary of how modeling was used in most of the situations. (see Application Case 4.8.) Most of the full articles contain additional information to flesh out the situation and add specifics that make the example come alive. This is less necessary in some of the other chapters where the topic, at least at the conceptual level, is easier for students to grasp. (The case studies suggested



under “Cases” in Section 4.14 can serve this purpose, but cases in the text have the advantage of already being available.)

ANSWERS TO END OF SECTION REVIEW QUESTIONS Section 4.1 Review Questions

1. What decision problem does the Chilean agency face?

The decision problem is how to get a maximum number of low-cost high-quality meals from vendors who have an incentive to reduce costs and inefficiencies. To solve this problem, they use auctions.

The auction process begins when JUNAEB contacts and registers potential vendors. The agency then evaluates the companies from a managerial, technical, and financial point of view and eliminates those that don’t meet minimum reliability standards. The decision problem also includes qualifying vendors according to two characteristics: (1) their financial and operating capacity and (2) their technical and managerial competence.

2. Would a decision process that does not use a model be able to accommodate the desired goals?

The answers will differ depending on student’s viewpoints. Students could argue that a model is needed to accommodate the desired goals, which are to maximize quality and quantity of meals and to minimize costs. A model represents the relationship (or represents a simplified relationship) of the variables in the decision situation. To determine the best solution or achieve the desired goals, the relationship among the variables has to be identified and understood.Conversely, students might claim that models are not necessary, but that the decision process would take longer without a model and not reach the optimal choice (outcome).

3. What other approaches could you use to make these decisions?

The agency could set cut-off points, for example, meal costs must be less than $2.00. Vendors that did not meet the cut-off points would be eliminated from consideration. Then the final decision could take into consideration another cut-off point. This approach could take longer and result in a sub-optimal decision.

4. What other applications of this type of auction model can you envision?

Any B2B procurement (e.g., buying office supplies, or raw materials) could use (and many do use) this auction model.




1. List three lessons learned from modeling.

This question refers to lessons found in the examples in this section, not to lessons learned from modeling in general. Many examples could be cited, among them these four: DuPont learned how different approaches to rail transportation would

work out. The University of Virginia Health Science Center (the institution studied

in the cited Rossetti and Selandari paper) learned how human couriers and mobile robots compare in making deliveries within a hospital.

Procter & Gamble learned the best shipping options from product sources to distribution centers.

American Airlines learned the optimum ascent and descent profiles for its aircraft.

2. List and describe the major issues in modeling.

Major modeling issues include: problem identification and environmental analysis: scanning the

environment to figure out what problems exist and can be solved via a model

variable identification: identifying the critical factors in a model and their relationships

forecasting: predicting the future use of multiple models: combining them to solve many parts of a complex

problem model categories: selecting the right type of model for the problem or sub-

problem model management: coordinating a firm’s models and their use knowledge-based modeling: how to take advantage of human knowledge

in modeling

3. What are the major types of models used in DSS?

The major types from Table 4.1 follow. Optimization with few alternatives Optimization via an algorithm Optimization via an analytical formula Simulation Heuristics (“rules of thumb”) Predictive models Other models



4. Why are models not used in industry as frequently as they should or could be?

There are several reasons. The most important is that useful models are often complex, requiring major investments of time and money as well as scarce, expensive expertise.

5. What are the current trends in modeling?

Trends in modeling are financial and risk simulation, predictive analytics, and optimization.


1. What is a decision variable?

A decision variable is a data element controlled by the decision maker, whose possible values describe alternative courses of action.

2. List and briefly discuss the three major components of linear programming.

Of the four components of any decision support mathematical model, linear programming uses result (outcome) variables, decision variables and uncontrollable variables (parameters). Linear programming models do not use the fourth component, intermediate result variables.There are other possible answers that do not tie in to the concepts of this section. For example, one could say that the three major components of a linear programming problem are its objective function, its decision variables, and its constraints. You might wish to connect this question to Section 4.8.

3. Explain the role of intermediate result variables.

An intermediate result variable is the result of one part of a model, which in turn is input to another part of the model. Its role is to simplify the modeling process by breaking a complex problem down into multiple simpler subproblems, to present data of management interest that might not otherwise be visible between system inputs and final outputs, and to aid in debugging a model by helping a model developer zero in on the cause of any difficulties.


1. Define what it means to perform decision making under assumed certainty, risk, and uncertainty.

Decision making under assumed certainty: the values of all variables affecting the decision, including future values, are known or can be assumed to be known.



Decision making under risk: the exact value of decision variables is not known, but their statistical probability distributions are known (or can be assumed to be).Decision making under uncertainty: even these distributions are not known.

2. How can decision-making problems under assumed certainty be handled?

Decision-making problems under assumed certainty can be handled by mathematical models that will yield an optimum solution.

3. How can decision-making problems under assumed uncertainty be handled?

Decision-making problems under uncertainty can be handled, first, by trying to get more information. If this is not possible or if the uncertainty remains after this has been done, it is necessary to evaluate the outcomes of many possible values of the decision variables. The choice then depends on the decision maker’s attitude toward risk. For example, one solution may have a higher expected value than another, but also a higher likelihood of a negative outcome. A manager with an inclination to gamble might choose that solution, whereas a more conservative and risk-averse manager would choose a solution that offers a lower probable return but also minimizes the likelihood of loss.

4. How can decision-making problems under assumed risk be handled?

Decision-making problems under assumed risk can be handled by taking the probability distributions of the decision variables and working through the model with those distributions. The result will be a distribution of outcomes.

Note that the mean of the distribution of outcomes is often not the outcome that would be obtained under assumed certainty if each variable had its mean value. This is because, in a complex model, decision variables interact in complex ways and effect is not always immediately obvious.

Another approach to problems under assumed risk is to select a random value from each decision variable’s probability distribution and to run the model, under assumed certainty, with those values. This is repeated until a statistical distribution of results is obtained. This approach is used when a model is too complex to work through with statistical distributions directly.


1. What is a spreadsheet?

“A program designed to perform general computational tasks using spatial relationships rather than sequential execution as their primary organizing principle.” (Wikipedia)

“A computer program in which figures arranged in the rows and columns of a grid can be manipulated and used in calculations.” (Oxford American Dictionary)



This section does not define the concept, relying on student’s experience to tell them what a spreadsheet is. Given the audience of this book, that should normally be appropriate. Instructors who assign this question with no further explanation should expect some students to search for definitions (finding one or more like the above two) whereas others come up with their own and still others use descriptive phrases from the text. If a specific approach is wanted, that should be stated in advance.

2. What is a spreadsheet add-in? How can add-ins help in DSS creation and use?

Spreadsheet add-ins are small programs designed to extend the capabilities of a spreadsheet package. They help in DSS creation and use because many add-ins are designed specifically for that purpose, often by structuring and solving specific types of models.

3. Explain why a spreadsheet is so conducive to the development of DSS.

A spreadsheet is conducive to DSS development because it provides an easily understood metaphor for the computation and typically incorporates many powerful modeling functions.


1. List and explain the assumptions involved in LP.

The LP approach involves both economic and technical assumptions.Economic assumptions are: Returns from different allocations can be measured by a common unit

(e.g., dollars, utility). The return from any allocation is independent of other allocations. The total return is the sum of the returns yielded by the different activities. All data are known with certainty. Resources are to be used in the most economical manner.



Technical assumptions are:

The objective function (that is to be maximized) is a linear combination of the outputs. Each output (decision variable) uses a linear combination of the inputs. Constraints on inputs are linear (or constant, which is a special case of linear) inequalities.

The requested explanations should show that the student understands each of the concepts, and is not simply parroting items from a list.

2. List and explain the characteristics of LP.

Linear programming is an optimization method used to solve problems in which the objective function and the constraints are all linear. It rests on the assumptions listed in the previous answer. LP problems can be solved by a wide variety of software for all popular computers.

3. Describe an allocation problem.

An allocation problem is a linear programming problem in which limited resources can be allocated among several possible uses, each of which yields a known return per unit and are subject to constraints.

4. Define the product-mix problem.

The product-mix problem is a linear programming problem in which a variety of different products are made from common resources. Each product requires a known resource mix and has a known profitability. Some resources are limited. Total profitability is to be maximized.

A product-mix problem can be viewed as an allocation problem. The difference is that a person formulating the problem as a product-mix problem is typically interested in the quantities of each product to be produced, while a person formulating the same problem as an allocation problem is usually interested in the quantities of each resource to be used. The solutions are identical, and each approach can produce both answers.

5. Define the blending problem.

The blending problem is a linear programming problem in which resources can be used in different ways to create a desired end product. The ways in which the resources combine to create the characteristics of the end product are known. Total cost of the end product is to be minimized.

6. List several common optimization models.

Several common optimization models are listed in the shaded box just before these review questions. These include the assignment problem, dynamic programming, goal programming, investment, linear and integer programming,



network models for planning and scheduling, replacement (capital budgeting), simple inventory models (e.g., economic order quantity) and transportation (minimizing cost of shipments, route optimization). A motivated student will be able to find others through a Web search.


1. List some difficulties that may arise when analyzing multiple goals.

It is usually difficult to obtain an explicit statement of the organization’s goals.

The importance of specific goals may change over time or in different situations.

Goals and subgoals are viewed and weighted differently by different people and in different parts of the organization. (Composite of multiple items from the list in the text.)

Goals change in response to changes in the organization and its environment.

The relationship between alternatives and their role in determining goals may be difficult to quantify.

Participants access the importance (priorities) of the various goals differently;

2. List the reasons for performing sensitivity analysis.

Sensitivity analysis attempts to assess the impact of a change in input data or parameters on the result variable(s). Reasons for using it listed in this section include: Revising models to eliminate too-large sensitivities Adding details about sensitive variables or scenarios Obtaining better estimates of sensitive external variables Altering a real-world system to reduce actual sensitivities Accepting and using the sensitive (and hence vulnerable) real world,

leading to the continuous and close monitoring of actual results

Another reason for using sensitivity analysis is to understand the sensitivity of different decision choices to variations in external conditions. One alternative may promise excellent results under nominal conditions but performs poorly if conditions vary from the nominal. Another may not turn out quite as well if all goes according to plan but will perform better if conditions vary from it. For example, one possible supplier may offer low prices for deliveries according to a pre-planned schedule but impose large penalties if the schedule is changed; another might have a higher base price but lower penalties for changing the delivery schedule.



3. Explain why a manager might perform what-if analysis.

A manager might perform what-if analysis to find out what will happen if a particular action is taken. A manager might also perform what-if analysis to try out several alternatives, choosing the one that works out best.

4. Explain why a manager might use goal seeking.

A manager might use goal seeking to find values of decision variables that enable him or her to meet a predetermined criterion. For example, a manager may need to find an advertising budget that will reach X households at a cost of not over $Y.

A drawback is that goal seeking can encourage “satisficing,” discussed in Chapter 2. A manager who finds an advertising program that meets the above objectives may not be motivated to find one that reaches more households for the given cost, or the required number at less cost.


1. What is a decision table?

A decision table is a way to organize information in a systematic way to prepare it for analysis. Typically, decisions are shown along one axis, states of nature on the other. The range of outcomes that can result from any decision, given the possible states of nature, can then be easily seen in one column (or row).

2. What is a decision tree?

A decision tree is an alternative representation of a decision situation, in which choices and states of nature are shown as alternating nodes along the branches of a tree. The range of outcomes that can result from any decision, given the possible states of nature, can then be easily seen by following all branches from that decision to their ends.

3. How can a decision tree be used in decision making?

By showing the decision maker the possible outcomes that could result from a given choice, the tree gives the decision maker information by which to compare choices.

4. Describe what it means to have multiple goals.

Having multiple goals means that a decision maker hopes to obtain the best possible combination of several factors, all of which depend on the decision to be made. For example, a student may want to find an instructor who is entertaining, has a good reputation for teaching, grades easily, assigns little homework, and



whose section meets at convenient times. It will be difficult, if not impossible, to achieve all these at the same time. Multiple goals require willingness to compromise.

(You may want to defer this question until students have read Section 4.9, or to refer them forward to it at this point.)


1. What is Analytic Hierarchy Process?

The analytic hierarchy process (AHP), is an excellent modeling structure for representing multicriteria (multiple goals, multiple objectives) problems—with sets of criteria and alternatives (choices)—commonly found in business environments..

2. What steps are needed in applying AHP?

The decision maker uses AHP to decompose a decision making problem into relevant criteria and alternatives. The AHP separates the analysis of the criteria from the alternatives, which helps the decision maker to focus on small, manageable portions of the problem

3. What are the differences between AHP software and software such as ChoiceAnalyst?

ChoiceAnalyst takes a direct approach, using a gradient-free derivative and probability and statistics to generate a sampling using a confidence level and confidence interval (whose default is a model option) and a ranking calculation. A global minimum is iteratively found that best satisfies the decision maker and criteria weights along with the paired preference weight matrix. The number of iterations to reach convergence depends on the particular model involved.


1. What is a search approach?

A search approach is a general method, or a category of closely related methods, used to find the best solution to a problem based on some criterion. The criterion usually involves maximizing or minimizing a numeric value.

2. List the different problem-solving search methods.

Analytical techniques, algorithms, blind searching, and heuristic searching

3. What are the practical limits to blind searching?

Practical limits are the time and computer resources available for the search.



4. How are algorithms and heuristic search methods similar? How are they different?

The similarity is that they are both search approaches by the above definition (see Q. 1 above).

The difference is that algorithms always find a solution, if one exists within the constraints of the method, and that solution can always be proven to be mathematically optimal. Heuristic search methods may or may not find one. If one is found, it may or may not be “best” by the stated criteria. Their advantages over algorithms are that they can be used for some problems that lack applicable algorithms, and they are often faster.

(This answer uses the term “algorithm” within the context of solving business problems. The term has a more general meaning: “an explicit numerical step-by-step procedure that produces a solution to a known problem.” Some algorithms do not involve optimization, as an algorithm to calculate square roots. Others apply to areas where no mathematical optimum exists, for example, Google’s search algorithm mentioned in the text. There is no way to know if the pages it finds are best for a particular purpose, since people will disagree on what “best” means in this context. A person searching for “diamonds,” wanting information on gemstones, might find its answers useful, but the same results would be less useful to someone in need of information on baseball parks, geometric shapes, or the suit in a deck of cards.)


1. List the characteristics of simulation.

An imitation of reality rather than a representation of it A technique for conducting experiments Fewer simplifications than with most other methods Descriptive rather than normative; users can use simulation results to

search if desired Usually used for problems too complex for numerical optimization

methods

2. List the advantages and disadvantages of simulation.

Advantages: The theory is fairly straightforward. Time can be compressed a great deal, quickly giving a manager some feel

as to the long-term effects of many policies. Simulation is descriptive rather than normative. This allows a manager to

pose what-if questions. Managers can use trial-and-error quickly, at little expense, accurately, and with low risk.

A manager can experiment to determine which decision variables and which parts of the environment are really important, and with different alternatives.



An accurate simulation model requires intimate knowledge of the problem, thus forcing its builder to interact with the manager. This leads to better understanding of the problem and the potential decisions available.

The model is built from the manager’s perspective. No generalized understanding is required of the manager; as every

component in the model corresponds to part of the real system. Simulation can handle a wide variety of problem types, such as inventory

and staffing, as well as higher-level managerial functions, such as long-range planning.

Simulation can include most real complexities of problems; simplifications are not needed. For example, it can use real probability distributions rather than approximate theoretical ones.

Simulation automatically produces many important performance measures.

Simulation is often the only DSS modeling method that can readily handle relatively unstructured problems.

Simulation generally can include the real complexities of problems; simplifications are not necessary.

Simulation automatically produces many important performance measures.

Simulation is often the only DSS modeling method that can readily handle relatively unstructured problems.

There are some relatively easy-to-use simulation packages. These include add-in spreadsheet packages, influence diagram software, Java-based (and other Web development) packages, and visual interactive simulation systems.

Disadvantages: An optimal solution cannot be guaranteed, though relatively good ones are

generally found. Simulation model construction can be a slow and costly process, although

newer modeling systems are easier to use than ever. Solutions and inferences from a simulation study are usually not

transferable to other problems because the model incorporates unique problem factors. It doesn’t generalize.

Simulation is sometimes so easy to explain to managers that analytic methods are often overlooked. (This is not a disadvantage of the method, but a practical caution about using it.)

Simulation software sometimes requires special skills because of its complexity. Some simulation packages require the model developer to be, in effect, a programmer.

3. List and describe the steps in the methodology of simulation.

The steps are shown in Figure 4.11. They are: Define the problem Construct the simulation model



Test and validate the model Design the experiments Conduct the experiments Evaluate the results Implement the results

4. List and describe the types of simulation.

Important types of simulation include probabilistic simulation, time-dependent and time-independent simulation, object-oriented simulation, and visual simulation.

Probabilistic simulation is simulation where one or more of the independent variables is described by a statistical distribution.

Time-dependent and time-independent simulations are distinguished by whether or not it is important, in order to model the system correctly, to know exactly when events occur.

Object-oriented simulation refers to simulation models built using object-oriented programming methods. This is more a function of the tools used than of the simulation model itself. (If your students are interested in programming, you can point out here that much pioneering work in object-oriented programming was done in developing simulation languages such as Simula.)

Visual simulation is discussed in the next section and defined in the answer to Review Question 1.


1. Define visual simulation and compare it to conventional simulation.

Visual simulation uses graphical representation to show the situation to the end user. It does everything a conventional simulation does, which is any technique for conducting experiments (such as what-if analyses) with a digital computer on a model of a management system, but does it using visually pleasing and informative representation.

Visual interactive simulation is a type of visual simulation in which input is provided directly to the visual representation of the system, with results visible immediately or nearly so.

2. Describe the features of VIS (i.e., VIM) that make it attractive for decision

makers.

The main attraction of VIS (VIM) is the graphical display. This type of interaction can help managers learn about the decision making situation, though some people



respond to graphical displays better than others. Non-technical managers, who may not understand a situation from data in tables or graphs, can grasp what goes on in a visual presentation easily.

3. How can VIS be used in operations management?

According to the text, VIS (VIM) “has been used in several operations management decisions.” Examples given there include plant operations and waiting lines. The two are related, as waiting line buildup indicates an unbalanced plant. With VIS, a manager can see work in progress piling up at one stage of a process—which is far less expensive than watching it pile up in the real factory!

4. How is an animated film like a VIS application?

An animated film is (with modern animation techniques; this question is not about hand-drawn animations of a few decades ago) a computer-generated representation of a physical system, as is a VIS application. While some animated films use computers simply to generate final artwork and to fill in the gaps between key frames created manually, others use complex algorithms to simulate realistic motion of animated people or animals, calculate trajectories of physical objects and determine what happens when two objects interact (e.g., a vehicle hitting a wall). These are true VIS applications, even if their purpose is not to support management decision making.


1. Explain how a spreadsheet system such as Excel can be used in DSS modeling.

Section 4.5, which students will have read by now, is devoted to this topic. This section points out that Excel includes many built-in models. It can thus produce the needed results in many situations as long as the required data can be entered or imported into a spreadsheet.

2. Explain why OLAP is a kind of modeling system.

OLAP is a flexible approach to data analysis. OLAP packages have evolved from a wide range of origins. Since analyzing data often requires models, OLAP software vendors have added modeling capabilities to their systems, making OLAP a modeling system in addition to its other features.

3. Identify three classes of models and list two kinds of problems each can solve.

Three classes of models mentioned in this chapter are quantitative software packages, the Excel spreadsheet system, and OLAP systems.

Quantitative software packages can solve general optimization problems as well as problems that arise in specific applications. Section 4.7 (The Structure of



Mathematical Models for Decision Support), 4.8 (Mathematical Programming Optimization, which includes linear programming) and 4.10 (Problem-Solving Search Methods) contain many examples.

The kinds of problems Excel can solve are discussed at length in Section 4.5.

OLAP systems can address (though perhaps not solve) any problem whose answer lies in a large data warehouse. These tend to be problems where historical data can serve as a guide to the future. Two examples are (a) airline yield management, where historical data indicate what fraction of the seats that will ultimately be sold on a given flight will be sold X days before it leaves; and (b) retail demand analysis, where it can indicate how sales break down by any set of factors and what the trends are in this.

4. List the reasons model management is difficult.

The development of generic model base management systems is difficult because there are so many different model classes that can be used. It is hard to know the ones to include and the ones that may need to be modified for use. Further, the selection of an appropriate model requires expertise because several models may be applicable to any one situation. Finally, each organization uses models differently, unlike database management systems, where most organizations’ databases follow the row-and-column relational model.


web viewteradata university and other hands-on exercises. team assignments and role-playing...

Documents