Beslissingsmodellen voor Logistiek Management LM04

1
Management ScienceClass 2 Decision Analysis 1
MBA, Term 1, 2003/2004Dr. Raf Jans Dr.
Moritz Fleischmannoffice F2-53 office
F1-38phone [phone] phone [phone] e-mail
rjans_at_fbk.eur.nl e-mail MFleischmann_at_fbk.eur.nl
Rotterdam School of Management
2
This course is about

• Decision making in a structured way using
  quantitative modeling techniques
• Excel is among the most powerful and versatile
  quantitative tools available to managers
• Why are decisions hard?
• What is a good decision?
• Topic of today Decision (Tree) Analysis

3
Case discussion the rules of the game

• Prepare
• Rearrange facts and interpret them
• Come to the class with a point of view
• Participate
• Introduction of the problem and context
• Balance between the focus and flow
• Balance between overcontrol and chaos
• Decision must be defended
• Different views are valuable
• Adapt
• Listen to other arguments
• Learn from each other

4
(No Transcript)
5
Introducing the case

• Freemark Abbey is located in the Napa valley,
  California.
• Produces about 38.000 cases of premium wine
• 1000 cases of Riesling wine
• Dry (20 sugar)
• Sweet (25 sugar)
• Botrytis (up to 35 sugar)
• Analyse Jaegers decision problem
• How would you solve the problem?
• What decision do you recommend?

6
Objectives

• Provide an introduction into the concepts and
  methodologies of Decision Analysis.
• Develop analytical skills to structure and solve
  problems using decision trees and analyse the
  solutions.
• Develop practical skills in solving these
  problems with decision support tools (Excel and
  Precision Tree).
• Give an insight in the application of Decision
  Analysis to business problems.

7
Decision Analysis

• Framework and methodology for rational decision
  making under uncertainty
• Structure overall problem as a sequence of
  decisions and events
• Identify
• Alternatives / options
• Objective
• Uncertainties events and probabilities
• Consequences
• Sequence

8
Decision Trees

• Graphical tool for structuring and analyzing
  decision making under uncertainty
• Describe decision problem by tree-like
  structurewith two types of nodes

end node
event node
decision node
probability
value
value
9
Decision Trees
decision node
TRUE / FALSE
Decision 1
payoff 1
Name decision Value of optimal decision
TRUE / FALSE
Decision 2
payoff 2
10
Decision Trees
event node
Probability 1
Event 1
payoff 1
Name uncertainty Expected value
Probability 2
Event 2
payoff 2
11
Decision Trees
End node
Chance of occurrence in optimal
solution Cumulative payoff of path
12
Decision Trees (contd.)

• Roll-back or Fold-back the tree to select
• decisions and evaluate the overall project
• evaluate each of the trees leaves as the sum
  of the values along the path leading to it
• evaluate each event node as the expected value
  across all possible outcomes
• evaluate each decision node by picking the
  best-valued alternative

13
Limitations

• Applicable for a moderate number of decisions and
  events
• Otherwise the size of the tree explodes, making
  it cumbersome to handle and hard to capture
• Specific techniques available for larger problem
  instances (dynamic programming,
  branch-and-bound,)

14
Applications

• Product development
• Power trading in electricity markets
• Portfolio management (pharmaceuticals)
• Location decisions (nuclear plant, airport,)
• Investment projects
• Medical diagnosis
• Oil exploration (drill or not?)
• Marketing (new product introduction)

15
The Value of Decision Analysis at Eastman Kodak
Company

• Because of the one-time nature of typical
  decision-analysis projects, organizations often
  have difficulty identifying and documenting their
  value. Based on Eastman Kodak Company s records
  for 1990 to 1999, we estimated that decision
  analysis contributed around a billion dollars to
  the organization over this time. The data also
  reflect the many roles decision analysis can
  play. Aside from its monetary benefits, it
  promotes careful thinking about strategies and
  alternatives, improved understanding and
  appreciation of risk, and use of systematic
  decision-making principles.
• Interfaces, Vol. 31 (5), Sep-Oct 2001, 74-92

16
How Bayer Makes Decisions to Develop New Drugs

• Drug development is time consuming, resource
  intensive, risky, and heavily regulated. To
  ensure that it makes the best drug-development
  decisions, Bayer Pharmaceuticals (Pharma) uses a
  structured process based on the principles of
  decision analysis to evaluate the technical
  feasibility and market potential of its new
  drugs. In July 1999, the biological products
  leadership committee composed of the senior
  managers within Bayer Biological Products (BP), a
  business unit of Pharma, made its newly formed
  strategic-planning department responsible for the
  commercial evaluation of a new blood-clot-busting
  drug. Pharma senior managers considered our
  recommendations relevant to their decision
  making. The project also institutionalized
  decision analysis at the business-unit level.
• Interfaces, Vol. 32 (6), Nov-Dec 2002, 77-90

17
Management and Application of Decision and Risk
Analysis in Du Pont

• Decision and risk analysis (DRA) enables Du
  Pont's business teams to develop creative
  strategy alternatives, evaluate them rigorously,
  select those with the greatest expected
  shareholder value, and design implementation
  plans the businesses can enthusiastically
  support. Du Pont organized internal and external
  resources to develop its DRA capability and
  incorporated DRA in several ongoing business
  processes. One Du Pont business utilized DRA
  techniques to develop a business strategy that
  enhances value by 175 million. 
• Interfaces, Vol. 32 (6), Nov-Dec 2002, 77-90

18
Software

• Student Version of DecisionTools Suite
• PrecisionTree (decision analysis)
• _at_Risk (simulation)
• Install the software on your laptop withthe
  CD-rom
• Folder Palisade DecisionTools
• Setup.exe
• Valid initially for 30 days
• Get authorization code (via the web) to obtaina
  1 year license

19
Implementation in PrecisionTree

• Demonstrate the implementation of the Freemark
  Abbey case using PrecisionTree
• Advantages of PrecisionTree
• Intuitive and easy to learn
• Fully integrated within a spreadsheet model
• Generate customized reports and graphs (risk
  profiles and sensitivity analysis)

20
Building the tree
21
Building the tree
22
Building the tree
23
Building the tree
24
Building the tree
25
Building the tree
26
Building the tree
27
Building the tree
28
Building the tree
29
Building the tree
30
Risk Profiles

• Each decision/strategy is linked to a set of
  potential results with associated probabilities
  gt risk profile
• Expected value alone provides limited
  informationgt What if alternative decision leads
  to slightly lower expected payoff but at a much
  lower risk?
• Expected value criterion assumes risk neutrality
• Risk is associated with a decision/strategy,not
  with a problem

31
Risk Profiles Freemark Abbey
32
Summary Decision Tree Methodology

• Structure the problem as a sequence of decisions
  and events
• Make sure overall sequence is correct
• Associate probabilities with random outcomes
• Associate payoffs with decisions and with random
  outcomes
• Roll-back the tree to select decisions and
  evaluate the outcome
• Events compute expected value
• Decisions pick best option

33
Summary Decision Tree Methodology

• Analyze risk profiles of different decisions
• Look beyond expected values
• Perform sensitivity analysis
• When do decisions change?
• Break-even analysis

34
Key Insights

• Uncertainty does not inhibit rational decision
  making
• Structure the problem you want to analyze
• Think carefully about
• Your objectives
• Your options
• Uncertainties events and likelihoods
• Consequences costs or payoffs
• Sequence
• Decision trees provide intuitive tool for
  sequential decision making under uncertainty

35
Key Insights

• Each decision corresponds with a specific risk
  profile
• Make a clear distinction between the quality of a
  decision and the quality of its outcome
• Analyze impact of parameter changes on decisions
  and outcomes
• Identify key drivers of a decision
• DA is a way of communicating your reasoning and
  analysis in a structured way

36
Outlook

• Class 4 Freemark Abbey revisited,advanced
  topics in decision analysis
• Preparation
• Browse through Chapter 10 Decision making under
  uncertainty of Winston Albright
• Think about the following questions related to
  the Freemark Abbey case
• Should Jaeger buy the Botrytis spores?
• Should Jaeger rent the Super Doppler?

37
Information on the web

• Decision Analysis Society homepage
  http//faculty.fuqua.duke.edu/daweb
• Software companies
• Palisade (http//www.palisade.com)
• Consultancy companies
• Strategic Decisions Group (http//www.sdg.com)
• Decision Strategies (http//decisionstrategies.com
  )
• ...

38
If you want to know more

• Application examples
• Various applications Interfaces
  http//www.interfaces.smeal.psu.edu/issues/specia
  l.php
• Volume 22, nr. 6, Nov-Dec 1992
• Volume 29, nr. 6, Nov-Dec 1999
• Medical decision making (Interfaces, Vol.28,
  Nr.4)http//www.interfaces.smeal.psu.edu/issues/r
  egular.php?article_idv28n4a8
• Managing hydropower in Brazil (OR/MS Today,
  April 2000) http//lionhrtpub.com/orms/orms-4-00/e
  scudero.html
• Software Survey (OR/MS Today, June 2002)
• http//lionhrtpub.com/orms/surveys/das/das.html

39
The end