Transcript Document
Defining Procedures for Decision Analysis May 02-14 & Engr 466-02A April 30, 2002 Client & Faculty Advisors – Dr. Keith Adams – Dr. John Lamont – Dr. Ralph Patterson III Team Members – – – – – – Marvin Choo Dave Cohen Amy Kalbacken Natasha Khan Jesse Smith Theodore Scott Acknowledgments Faculty Advisors Dr. Doug Gemmil Dr. Kenneth Kirkland Dr. Jo Min Dr. Ron Nelson Dr. Steve Russell Dr. Howard Van Aucken Dr. Max Wortman Presentation Outline Problem Statement Design Objectives End-Product Description Assumptions & Limitations Project Risks & Concerns Technical Approach Evaluation of Project Success Presentation Outline Recommendations for Future Work Personnel Budgets Financial Budgets Lessons Learned Closing Summary Problem Statement Problem – Companies often are required to make major decisions regarding the commercialization process for a product, process, or service – How can we maximize efforts most efficiently during the decision-making process? Goal – Develop a guide that aids users in the decisionmaking process Design Objectives Design Constraints – Inaccurate research (especially Internet) – Uncovering all factors – Limited understanding of algorithms Design Objectives Intended Users & Uses – People in decision-making positions Gain greater understanding of methods – Software Programmers Have background reference information Detailed starting point for developing software End Product Description The report will aid individuals in conducting a thorough analysis of the decision factors surrounding their particular product, process, or service End Product Description Written Report – Key factors regarding decision processes – Algorithms used in decision analysis – Examples of Algorithms – Functional Software Specification – Reference Material Assumptions Considering company goals Aids in decision-making but will not be the only tool used Take into account other decision-making factors and considerations Using decision-making algorithms Assumptions Use algorithms based on research Have basic knowledge of decision-making process For any business interested in decision analysis software No sophisticated mathematics or statistics are used in algorithms Limitations Ranking the importance of each factor differently Not all data accounted for Selected algorithms may not be applicable to all decisions Need to apply each process to specific situation Limitations Limited knowledge of algorithms Algorithms may require a statistical background or other expertise All factors & constraints may not be uncovered Algorithm applicability is based on project requirements & criteria Project Risks & Concerns Scheduling interviews Finding information Losing a team member Understanding project Technical Approach Purpose To determine an algorithm for use in creating software that will implement the decision analysis process Process Determine the basic project process Compile a list of potential algorithms Create a set of criteria for evaluating the algorithms Research the algorithms Select the most applicable algorithms Technical Approach “Basic Project Process” Technical Approach “List of Algorithms” Artificial Neural Networks Bayesian Logic Decision Matrix Decision Tree Fuzzy Logic Genetic Algorithms Linear Algebra Technical Approach “Criteria for Evaluating the Algorithms” What type of problems is the algorithm good for? What input data is needed? What kind of control is needed? How does the algorithm work? What are the expected outputs? How easy or difficult is it to implement? Is there any information on the solution time, problem size, etc. Are there any examples available for the algorithm? Are there sufficient conditions for convulgence? If the algorithm is discovered to be ineffective what are the reasons in support of the determination. Technical Approach “Selecting the Best Algorithms” Artificial neural networks Able to learn, memorize, and create relationship between data Able to work with the non-linearities Used for the accurate prediction of events Decision trees Useful for handling a lot of complex information Genetic algorithms Multi objective solutions can be defined Project Success Initial Startup – Identifying key factors – Interview coordination Interview Results & Project Definition – Conduction Interviews – Completing Project Plan Project Success Implementation – Algorithms – Functional Software Specification Testing – Scenario Example – Needs further testing Project Success End Product – Guide Algorithms Report Functional Software Specification Reference Material – Final Report Software Package Recommendations for Further Work Create detailed models of selected algorithms Consult with professionals to evaluate algorithms Develop a functional software package Personnel Budget Planned Revised Actual Dave Cohen 77 87 90 Amy Kalbacken Theodore Scott 88 95 96 75 86 90 Marvin Choo 81 81 85 Natasha Khan 79 79 83 Jesse Smith 89 89 93 Budgeted Hours Vs. Actual Hours Hours Personnel Effort 100 90 80 70 60 50 40 30 20 10 0 Budget To Date David Amy HongViet Theodore Marvin Cohen Kalbacken Nguyen Scott Choo Personnel Natasha Khan Jesse Smith Financial Budget Item Original Estimate Cost Printing $60.00 Transportation $0.00 Labor $0.00 Equipment & Parts $0.00 Telephone $0.00 Total Estimated Cost $0.00 Revised Estimate Cost Actual Cost to Date $45.00 $42.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 $0.00 Lessons Learned Essential team attributes – Teamwork – Time Management – Brainstorming Knowledge acquired – Division of labor, team goals, task management – Interview information – Defining scope of project Lessons Learned Algorithms – Complexity – Need to study more carefully Issues faced in decision-making process – Time vs. Money – Who is involved in decision-making process – Engineering vs. Business Processes Closing Summary Conclusion A tool created to aid during the decision-making process would be well worth developing Benefits Identifies key factors in the decision process Characterizes the decision-making process Determines the best decision processes Aid in further analyzing a particular decision Narrows in on the optimum decision Questions