EMBA – Decision Analysis Agenda 1. Interfaces Gawne, Aron Kirby, Mary Rutberg, Matthew Ananturi, Siv PLATO Helps Athens Win Gold: Olympic Games Knowledge Modeling for Organizational Change and Resource Management Research and Development Project Valuation and Licensing Negotiations at Phytopharm, PLC Pricing Analysis for Merrill Lynch Integrated Choice A Multimethod Approach for Creating New Business Models: The General Motors OnStar Project 2. Bidding a. Dom’s $20 game b. Howard’s $20 game 3. Homework review a. Transportation & Assignment b. Integer Programming 4. Decision Tables and results (Example 1) a. Expected Monetary Value (EMV) or Expected Value or Expected Profits b. Maximin (ignores probabilities) c. Maximax (ignores probabilities) d. Hurwicz α*best + (1-α)* worst (ignores probabilities) e. Regret i. Minimax regret ii. Minimum average regret f. Value of Perfect Information g. LaPlace (equal probabilities) h. Excel – Scenarios 5. Creating the decision table (Example 2) a. By hand (whiteboard) b. Excel QM 6. Decision tables as a decision tree a. Example 1 revisited - introducing Excel QM Decision Tree model 7. Decision trees (sequential) a. Example 3 – job hunting i. By hand ii. Excel QM iii. Example 3b b. Example 4 – investment 8. Note: “Optimal Sequential Decisions and the Content of the Fourth-and-Goal Conference by Hurley; Interfaces 28:6 Nov-Dec 1998 (pp19-22) 106768242 – Page 1 9. Excel Data Tables (time permitting) 106768242 – Page 2 Exercise 1: A decision table example. A company is considering modifying its capacity. There are four options available. 1. Overtime – this is very flexible but there is a limit to the amount of capacity which can be added 2. Subcontract – this is also flexible and limited 3. Hire part-timers – we can not hire and fire every day so this is not as flexible as either of the two previous options but allows us to add more capacity than just using overtime or subcontracting 4. Hire full timers – this is the least flexible but enables us to add the most capacity The future is uncertain. We are not sure if demand will be low, medium or high. The profits depend on the demand and our decision and are given in the table below. Alternative\Probability Overtime Subcontract Hire part time Hire full time 106768242 – Page 3 Low demand 0.2 5 3 -1 -7 Normal demand 0.5 7 6 9 10 High demand 0.3 10 12 14 15 Solution using Excel QM (or QM for Windows) Decision Tables Data Profit Probability Decrease 0.2 Overtime Subcontract Hire part time Hire full time Same 5 3 -1 -7 0.5 Increase 0.3 7 6 9 10 10 12 14 15 Results EMV Maximum Expected Value of Perfect Information Column best 5 10 7.5 7.2 8.5 8.1 8.5 15 10.5 8.5 2 Increase 0.3 Expected 0.5 3 4 1 0 5 3 1 0 3 3.3 2 2.4 2 Minimum 5 3 -1 -7 5 Decrease 0.2 Overtime Subcontract Hire part time Hire full time 0 2 6 12 Same Minimum Decision making rules Expected (monetary) value Maximum of the minima (maximin)/Worst case scenario analysis Maximum of the maxima (maximax)/Best Case analysis o Sports discussion Minimum of the maxima regrets (minimax regret) (Expected regret) Concept Expected value of perfect information Excel Scenarios 106768242 – Page 4 10 12 14 15 15 <-Expected value under certainty <-Best expected value <-Expected value of perfect information Regret Probability Maximum Maximum 5 4 6 12 4 Example 2: Creating the decision table. In August, a bookstore must decide how many of next year's nature calendars to order. Each calendar costs the bookstore $6.00 and is sold for $10.00. After February 1, all unsold calendars are returned to the publisher for a refund of $.50 per calendar. Based on past years the number of calendars that will be sold is distributed as follows: Number Sold 100 150 200 250 300 Probability .3 .2 .3 .15 .05 Create the decision table by hand Profit Order 100 Order 100 Order 150 Order 200 Order 250 Order 150 Order 200 Order 250 Order 300 Create the decision table using Excel QM – Single Period Inventory 106768242 – Page 5 Order 300 Decision Trees Create the decision table from example 1 as a decision tree using Excel QM Example 3: Sequential decision tree – job hunting Maurice Cheeks is currently employed at a salary of $100,000 but is unhappy with his job. While Maurice is unhappy due to lack of support from his manager his major concern is with salary. Maurice has taken two immediate actions in order to try and alleviate the problem. He has applied for a job with another firm (at a salary of $120,000) and he has applied for a new position in his current firm (at a salary of $140,000). As life has it, Maurice has been offered the new job with the new firm and has 10 days to decide whether or not to take the offer. Unfortunately, he will not find out about the internal transfer for at least a month. Maurice feels that if he stays with his current firm there is a 70% chance that he will get the new position. If he does not get the new position then he can, of course, continue in his current position or he is considering quitting his current job on principle in order that he may devote full time to searching for a new job (after a brief vacation in Europe). If he looks for a new job then he figures there is a 20% chance he will land a job with a salary of $130,000; a 40% chance that he will land a job with a salary of $110,000, a 30% chance that he will land a job with a salary of $90,000 and a 10% chance that he will not land a job and therefore have to perform menial labor earning $40,000. Should Maurice take the job offer in hand or not? Create the decision tree using Excel QM Example 3b: Suppose the job search costs $5,000. 106768242 – Page 6 Example 4: Sequential decision tree – investment An investor is deciding between purchasing an apartment building for $800,000 or purchasing land for $200,000. If the investor purchases the apartment building, either the population of the town will grow over the next 10 years (p=.6) and the (NPV) revenue will be $2,000,000 or the population will not grow over the next 10 years (p=.4) and the (NPV) revenue will be $225,000. Payoffs are over a 10 year period. If the investor chooses to purchase land, there is a 50% probability that there will be population growth over the next 3 years at which time a decision will need to be made about the development of the land. If there has been population growth the investor will consider building apartments costing $800,000 or selling the land for $450,000. If there has not been population growth the investor will consider developing the property commercially costing $600,000 or selling the land for $210,000. (Notice the land sells for less if there has been no growth rather than if there has been growth.) If the apartments are built and if there is population growth over the next 7 years (p=.8) then the (NPV) revenue will be $3,000,000. If the apartments are built and if there is no population growth over the next 7 years (p=.2) then the (NPV) revenue will be $700,000. If the land is developed commercially and if there is population growth over the next 7 years (p=.3) then the (NPV) revenue will be $2,300,000. If the land is developed commercially and if there is no population growth over the next 7 years (p=.7) then the (NPV) revenue will be $1,000,000. (Notice that the probabilities of growth over the last 7 years differ because in one case there was growth in the first 3 years while in the other case there was no growth in the first 3 years.) Create the decision tree using Excel QM 106768242 – Page 7 Homework Chapter 3, Page 103 1. Problems 16, 17, 18 - Either QM for Windows or Excel QM can be used for these problems. In addition to answering the questions be sure that you can explain ALL of the results that are presented. 2. Problem 25 - Model as a decision table and find all results available in the software. I suggest that you use Excel QM (rather than QM for Windows) in order that you may use Excel to create the table of data. 3. Problems 28 & 29 – Use Excel QM 4. Let’s Make a Deal There are two envelopes, each containing an amount of money; the amount of money is either $5, $10, $20, $40, $80 or $160 and everybody knows this. Furthermore, we are told that one envelope contains exactly twice as much money as the other. The two envelopes are shuffled, and we give one envelope to you and one to your opponent. After the envelopes are opened (but the amounts inside are kept private) you and your opponent are given the opportunity to switch. If you both want to switch, we let you. Find the optimal strategy for playing this game. A strategy identifies what you would do under all situations. Thus for this example you need to decide between switch and don't switch for any amount of money you are looking at. Amount of money in your envelope $5 $10 $20 $40 $80 $160 106768242 – Page 8 Decision (Circle one) Offer to switch, Do not offer to switch Offer to switch, Do not offer to switch Offer to switch, Do not offer to switch Offer to switch, Do not offer to switch Offer to switch, Do not offer to switch Offer to switch, Do not offer to switch