1. Review for exam II—fall 2013 October 16, 2013

2. Format for exam 30 -- 40 multiple choice 3 sets of discussion questions Identical in format to previous exam

3. Bring…/Don’t Bring… Bring… –Scantron sheet –Pencil, eraser, calculator Don’t Bring… –Paper –PDAs, Pocket PC’s, tablets, –Programmable, high memory storage devices

4. We Covered: Burns Chs 4, 7, 8 Schwalbe 3, 4, 6 {will not test you on Burns ch 8 on this exam}

5. Estimation Weakest link History database By analogy Use models Top down—uses…..XXXXXXX?? Bottom up – uses…..XXX?? Reconciliation of top down and bottom up

6. Parkinson’s Law What is it??

7. Processes

8. More Processes

9. What are the processes that make up the cost management knowledge area? Estimate Costs Determine Budget Control Costs

10. What are the processes that make up the quality management knowledge area? Plan Quality Perform Quality Assurance Perform Quality Control

11. Methodologies Waterfall – Document-driven—CASE tools Spiral—risk-driven—invented by Barry Bhoem RAD—superior to SAD Agile/Iterative—increasingly popular –Scrum –RUP

12. Process Map Waterfall Iterative Low Ceremony High Ceremony Little Doc, light process discipline Heavy Doc, heavy process discipline, CCB Risk Driven, Continuous integration and testing Few risks, late integration and testing

13. The Waterfall Staircase Definition of Requirements Design Acceptance Testing Implementation Operation

14. Figure 2-3. Spiral Model of Software Dev.

15. What is the homegound for Waterfall? Stable requirements – few changes Large monolithic app—can’t be broken up Difficult development –Mathematical algorithms –Compliers –Database engines –Artificial intelligence apps

16. Home Ground for Agile Unstable/unknown requirements Rapid technological change An environment accommodative of change An environment conducive to learning The latter two are accommodated together

17. The Transform Model Did it work?

18. We also covered Probabilistic PERT (formulas will be given to you) –Each task (activity) requires three time estimates – Optimistic, Most likely, Pessimistic Crashing

19. Further comments Burns 6-13: Almost all of you did this wrong—you did not follow my solution in class, nor did you follow the solution provided in Chapter 6 of Burns

20. Further Recitation What is meant by feasibility? What are three kinds of feasibility? –General, economic (financial), technical What is meant by a risky decision? –State probabilities are KNOWN –Upside only, downside only, or both What is meant by an uncertain decision? How can decision theory help us make better project decisions?

21. Know MS Project Navigation What tool is used to specify subordination? What tool is used to link tasks sequentially? What tool is used to assign resources? How do you enter resources and their hourly rates? How do you show COST on the Entry table? How do you specify durations? What enables you to see ES, EF, LS, LF, Slack

22. Topics--Chapter 6 Must understand what Bayesian revision does for you Will have to do analyses exactly like that in Chapter 6 –DMUU—Decision Making under Uncertainty –DMUR—Decision Making under Risk

23. Decision Making under Uncertainty Choose alternatives—must be mutually exclusive Choose states—must be mutually exclusive and collectively exhaustive For each alternative/state pair specify a payoff What are the decision criteria? They are based on the DM’s attitude toward the situation—pessimist, optimist, regretist, etc

24. To solve a problem like 6-10, you must Do both DMUU and DMUR Recall that EPPI and EVPI are not decision criteria, like EV* (maximal expected value) and ER* (minimal expected regret)

25. To solve a problem like 6-12 or 6- 13, you must First, solve the decision problem without any additional information –Determine the optimal choice –Compute EPPI and EVPI Second, perform Bayesian revision Third, solve decision problem assuming a positive predictor state (consultant predicts SUCCESS) Fourth, solve decision problem assuming a negative predictor state (consultant predicts FAILURE) Compute EPSI and EVSI

26. Remember… EPPI = expected payoff of perfect information is not a decision criterion … tells us the payoff of perfect information EVPI = EPPI – EV* (WITHOUT ADD’TL INFORMATION) … tells us how much perfect information would be valued EVPI = expected value of PERFECT information EPSI = expected payoff of sample information EVSI = EPSI – EV* (WITHOUT ADD’TL INFORMATION) EVSI = expected value of sample information… tells us how much sample information would be valued—how much we might be willing to pay for additional information

27. Decisions Are always embedded in a _____ Are affected by ______ How to get better MM’s? –Build mathematical and simulation models