Lecture 1 from MIT Communicating with Data Course
1 like531 views
The document outlines the course structure for a summer 2003 management program titled 'Communicating with Data', led by Professor John S. Carroll. It covers various aspects of decision-making, including analysis, judgment, and the use of decision trees to inform and communicate decisions effectively. The course emphasizes the importance of understanding data and its implications in management decisions, as well as assessing the framework for analyzing complex choices.
Lecture 1 from MIT Communicating with Data Course
- 1. 15.06315.063 CommunicatingCommunicating with Datawith Data Sloan Fellows/Management of TechnologySloan Fellows/Management of Technology Summer 2003Summer 2003
- 2. 15.063 Summer 200315.063 Summer 2003 22 IntroductionsIntroductions Prof.:Prof.: John S. CarrollJohn S. Carroll See master schedule forSee master schedule for lectures and recitationslectures and recitations Office hours forOffice hours for TI’sTI’s will be postedwill be posted Syllabus:Syllabus: this is our roadmap for the course. Pleasethis is our roadmap for the course. Please read it carefully. We try to follow it as closely as weread it carefully. We try to follow it as closely as we can. Let’s take a look…can. Let’s take a look…
- 3. 15.063 Summer 200315.063 Summer 2003 33 Course OutlineCourse Outline Course Philosophy and ApproachCourse Philosophy and Approach Decision TreesDecision Trees ProbabilityProbability –– Discrete and ContinuousDiscrete and Continuous SimulationSimulation RegressionRegression Decision Making Examples and ExercisesDecision Making Examples and Exercises Communicating with DataCommunicating with Data
- 4. 15.063 Summer 200315.063 Summer 2003 44 Course GradingCourse Grading Cases andCases and HomeworksHomeworks 40%40% Class ParticipationClass Participation 10%10% Final ExamFinal Exam 50%50% Assignments indicate Read, Prepare, orAssignments indicate Read, Prepare, or Hand inHand in Questions?Questions?
- 5. 15.063 Summer 200315.063 Summer 2003 55 Course PhilosophyCourse Philosophy Good managers are skilled decisionGood managers are skilled decision makersmakers Decision making requires information,Decision making requires information, analysis, judgment, and intuitionanalysis, judgment, and intuition Information is constructed from bits andInformation is constructed from bits and pieces of ambiguous datapieces of ambiguous data Judgment involves understanding orJudgment involves understanding or framing situations and clarifying valuesframing situations and clarifying values
- 6. 15.063 Summer 200315.063 Summer 2003 66 Communicating with DataCommunicating with Data We “communicate” with data byWe “communicate” with data by constructing information in order to makeconstructing information in order to make effective decisions and get resultseffective decisions and get results We also use data to communicate, to tell aWe also use data to communicate, to tell a persuasive story to stakeholderspersuasive story to stakeholders In the spirit of the MyersIn the spirit of the Myers--Briggs TypeBriggs Type Indicator, combine competency in analysisIndicator, combine competency in analysis and intuition, “seeing” and judging, dealingand intuition, “seeing” and judging, dealing with ideas, material resources, and peoplewith ideas, material resources, and people
- 7. 15.063 Summer 200315.063 Summer 2003 77 Analysis and JudgmentAnalysis and Judgment Data Information analysis Situation, context Decisions, strategies, actions judgment judgm ent Analysis informs judgment, builds intuitionAnalysis informs judgment, builds intuition Analysis is not a substitute for judgmentAnalysis is not a substitute for judgment
- 8. 15.063 Summer 200315.063 Summer 2003 88 What Does the Data Mean?What Does the Data Mean? During WWII, 10,000 US bombers were lost, andDuring WWII, 10,000 US bombers were lost, and many others returned to base with damagemany others returned to base with damage An analysis was done of the location of damage,An analysis was done of the location of damage, proposing to reinforce some areas of the planesproposing to reinforce some areas of the planes Battle-damaged B-17s, Courtesy of US Air Force
- 9. 15.063 Summer 200315.063 Summer 2003 99 Medical Decision ExampleMedical Decision Example 389 schoolboys screened by a panel of389 schoolboys screened by a panel of three doctors: 45% judged to need theirthree doctors: 45% judged to need their tonsils removedtonsils removed 215 who were judged215 who were judged notnot to need theirto need their tonsils removed were examined by a newtonsils removed were examined by a new panel of doctorspanel of doctors What % should be judged to need theirWhat % should be judged to need their tonsils removed?tonsils removed?
- 10. 15.063 Summer 200315.063 Summer 2003 1010 Medical Decision, ContinuedMedical Decision, Continued Results: 46%Results: 46% 116 boys judged twice not to need their116 boys judged twice not to need their tonsils out were judged by a new paneltonsils out were judged by a new panel of three doctorsof three doctors Results: 44%Results: 44%
- 11. 15.063 Summer 200315.063 Summer 2003 1111 How Did Doctors Decide?How Did Doctors Decide? ExperienceExperience –– In the past, a bit less than half of patientsIn the past, a bit less than half of patients presenting themselves had their tonsils outpresenting themselves had their tonsils out –– Minimal systematic feedback: years ofMinimal systematic feedback: years of experience improve technical skills, but notexperience improve technical skills, but not decision making behaviordecision making behavior Relative JudgmentsRelative Judgments –– Who had the bigger, redder tonsils?Who had the bigger, redder tonsils? –– Much easier than yes/no absolute judgmentsMuch easier than yes/no absolute judgments
- 12. 15.063 Summer 200315.063 Summer 2003 1212 How Should They Decide?How Should They Decide? How can we structure the decision?How can we structure the decision? What are theWhat are the GGoals/values associated withoals/values associated with the outcomes?the outcomes? What are theWhat are the OOptions/action alternatives?ptions/action alternatives? What are theWhat are the OOutcomes?utcomes? What are theWhat are the PProbabilities/uncertainties?robabilities/uncertainties? How can we use the analysis to inform ourHow can we use the analysis to inform our decision making?decision making?
- 13. 15.063 Summer 200315.063 Summer 2003 1313 Decision AnalysisDecision Analysis Decision trees provide an elegant framework forDecision trees provide an elegant framework for combining options, contingencies, consequencecombining options, contingencies, consequence probabilities, and outcome values to help youprobabilities, and outcome values to help you select the best option.select the best option. Decision trees map all options and potentialDecision trees map all options and potential consequences in a manner that makes it easierconsequences in a manner that makes it easier to understand and communicate the situation.to understand and communicate the situation.
- 14. 15.063 Summer 200315.063 Summer 2003 1414 Decision TreesDecision Trees List options (include all possible action alternatives!)List options (include all possible action alternatives!) List uncertain events (mutually exclusive and collectivelyList uncertain events (mutually exclusive and collectively exhaustive)exhaustive) Construct a decision tree along a time line:Construct a decision tree along a time line: –– decision nodesdecision nodes (list choices)(list choices) –– event nodesevent nodes (list events)(list events) Evaluate endpoints (outcomes for each end branch)Evaluate endpoints (outcomes for each end branch) Assess event probabilitiesAssess event probabilities “Expect“Expect--out and Fold Back” = Backwards Inductionout and Fold Back” = Backwards Induction Sensitivity AnalysisSensitivity Analysis What does it mean for decision making?What does it mean for decision making?
- 15. 15.063 Summer 200315.063 Summer 2003 1515 Remove Tonsils?Remove Tonsils? remove leave sick not sick apologize don’t tell lawsuit sick no lawsuit not sick
- 16. 15.063 Summer 200315.063 Summer 2003 1616 Analysis Paralysis?Analysis Paralysis? The tonsillectomy decision tree could get veryThe tonsillectomy decision tree could get very “bushy” (complex), ambiguous, time consuming“bushy” (complex), ambiguous, time consuming Many possible contingencies and uncertaintiesMany possible contingencies and uncertainties Therefore, doctors don’t analyze this wayTherefore, doctors don’t analyze this way Instead, medical practice and research createsInstead, medical practice and research creates simpler decision rules (heuristics)simpler decision rules (heuristics) Imagine a researchImagine a research--based guide with pictures ofbased guide with pictures of tonsils: best medical practice would match thetonsils: best medical practice would match the picture and follow the guide unless there is apicture and follow the guide unless there is a reason to override (new decision analysis!)reason to override (new decision analysis!)
- 17. 15.063 Summer 200315.063 Summer 2003 1717 WellWell--Structured DecisionsStructured Decisions Known list of action alternativesKnown list of action alternatives Measurable outcomes, often monetaryMeasurable outcomes, often monetary Uncertainties can be stated as probabilityUncertainties can be stated as probability or probability rangeor probability range
- 18. 15.063 Summer 200315.063 Summer 2003 1818 Decision Analysis SkillsDecision Analysis Skills The skills of decision analysis are not in theThe skills of decision analysis are not in the computationscomputations The skills are in applying these concepts to aThe skills are in applying these concepts to a wider range of real decisionswider range of real decisions The decision tree calculations/sensitivityThe decision tree calculations/sensitivity analysis can be implemented in Excel as shownanalysis can be implemented in Excel as shown in the text, and the actual decision tree can bein the text, and the actual decision tree can be drawn usingdrawn using TreePlanTreePlan. Different versions are. Different versions are available at:available at: http://http://www.treeplan.com/treeplan.htmwww.treeplan.com/treeplan.htm
- 19. 15.063 Summer 200315.063 Summer 2003 1919 Closing CommentsClosing Comments ““Commercial Strength” alternatives:Commercial Strength” alternatives: DecisionProDecisionPro ($795 14($795 14--day free demo available)day free demo available) http://www.vanguardsw.com/default.htmhttp://www.vanguardsw.com/default.htm PrecisionTree ProPrecisionTree Pro ($795 Excel Compatible)($795 Excel Compatible) http://www.palisade.com/html/ptree.htmlhttp://www.palisade.com/html/ptree.html Be ready for lecture 2: study chapter 1,Be ready for lecture 2: study chapter 1, read 2.1, 2.2., 2.3 and prepare theread 2.1, 2.2., 2.3 and prepare the KendallKendall Crab and Lobster Inc.Crab and Lobster Inc. Case.Case.