Decision Tree Analysis

32
Decision Analysis 5 th Apr 2014

Transcript of Decision Tree Analysis

Page 1: Decision Tree Analysis

Decision Analysis

5th Apr 2014

Page 2: Decision Tree Analysis

Agenda

1. Objective

2. Literature Review

3. Decision making

Overview

Decision making Environment

Decision Making Criteria

4. Models

5. Case

6. References

2

Page 3: Decision Tree Analysis

Objective

How the Decision Analysis can help in

decision making in the face of uncertainty

3

Page 4: Decision Tree Analysis

Literature Review

Decision analysis provides a framework and

methodology for rational decision making

when the outcomes are uncertain.

4

Page 5: Decision Tree Analysis

WCB

Worker’s Compensation Board of British

Columbia, Canada

Over 1,65,000 employers

1.8 million workers

Spends US$1 billion p.a.

Objective – Improve service & Reduce cost

5

Page 6: Decision Tree Analysis

WCB

Applying Decision Analysis with decision trees WCB is now saving approximately US $4

million per year while also enabling some injured workers return to work sooner

Source: Ernest Urbanovich, Ella E. Young, Martin L. Puterman, Sidney O. Fattedad, (2003) Early Detection of High-Risk Claims at the

Workers' Compensation Board of British Columbia. Interfaces 33(4):15-26.

http://dx.doi.org/10.1287/inte.33.4.15.16372 6

Page 7: Decision Tree Analysis

Westinghouse

Westinghouse Science and Technology

Center

R&D Arm to develop new technology

Objective – Deliver high impact technology

quickly & Reduce cost

7

Page 8: Decision Tree Analysis

Westinghouse

OR team developed a decision tree approach to analyzing any R&D proposal while considering its complete sequence of key decision points.

A decision tree with a progression of decision nodes and intervening event nodes provided a natural way of depicting and analyzing such an R&D project.

Source: Robert K. Perdue, William J. McAllister, Peter V. King, Bruce G. Berkey, (1999) Valuation of R and D Projects Using Options

Pricing and Decision Analysis Models. Interfaces 29(6):57-74.

http://dx.doi.org/10.1287/inte.29.6.578

Page 9: Decision Tree Analysis

ConocoPhillips

Conoco Inc. and Phillips Petroleum

company

3rd largest integrated energy co. in US

Objective – Judicious Allocation of

investment capital across a set of

exploration projects

9

Page 10: Decision Tree Analysis

Westinghouse

Source: Michael R. Walls, G. Thomas Morahan, James S. Dyer, (1995) Decision Analysis of Exploration Opportunities in the

Onshore US at Phillips Petroleum Company. Interfaces 25(6):39-56.

http://dx.doi.org/10.1287/inte.25.6.3910

In early 1990’s – Industry leader

in application of OR

methodology

DISCOVERY – decision

analysis s/w package

• Evaluate exploration projects

• Rank Projects

• Budget Consideration

Page 11: Decision Tree Analysis

Should You Ask?

Sir, why is my coursework marks so low? I deserve higher marks. Hehehe!

Page 12: Decision Tree Analysis

Which Mobile Phone should I buy?

What are the things you consider before making a decision?

Page 13: Decision Tree Analysis

Whom should I marry?

What are the things you consider before making a decision?

Page 14: Decision Tree Analysis

Decision

14

A general approach to decisionmaking that is suitable to a widerange of operations managementdecisions:

Capacity planning

Product and service

design

Equipment selection

Location planning

Page 15: Decision Tree Analysis

15

Decision Making Overview

Decision Making

Certainty Nonprobabilistic

Uncertainty Probabilistic

Decision Environment Decision Criteria

Page 16: Decision Tree Analysis

16

The Decision Environment

Certainty

Uncertainty

Decision Environment Certainty: The results of decision

alternatives are known

Example:

Must print 10,000 color brochures

Offset press A: $2,000 fixed cost

+ $.24 per page

Offset press B: $3,000 fixed cost

+ $.12 per page

*

Page 17: Decision Tree Analysis

17

The Decision Environment

Uncertainty

Certainty

Decision Environment

Uncertainty: The outcome that will occur

after a choice is unknown

Example:

You must decide to buy an item now or wait.

If you buy now the price is $2,000. If you

wait the price may drop to $1,500 or rise to

$2,200. There also may be a new model

available later with better features.

*

(continued)

Page 18: Decision Tree Analysis

18

Decision Criteria

Nonprobabilistic

Probabilistic

Decision CriteriaNonprobabilistic Decision Criteria: Decision

rules that can be applied if the probabilities of

uncertain events are not known.

* maximax criterion

maximin criterion

minimax regret criterion

Page 19: Decision Tree Analysis

19

Nonprobabilistic

Probabilistic

Decision Criteria

*

Probabilistic Decision Criteria: Consider the

probabilities of uncertain events and select

an alternative to maximize the expected

payoff of minimize the expected loss

maximize expected value

minimize expected opportunity loss

Decision Criteria

(continued)

Page 20: Decision Tree Analysis

Step 1

Identify

possible

future

conditions

or state of

nature

Develop a

list of

possible

alternatives

Determine

the payoff

associated

with each

alternative

for every

possible

future

condition

Estimate

the

likelihood of

each

possible

future

conditions

Evaluate

alternatives

based to

some

decision

criterion,

and select

the best

alternative

Decision Making Process:

Step 5Step 4Step 3Step 2

Page 21: Decision Tree Analysis

21

Decision Models

Decision Models

Payoff Matrix

Decision Tree

Page 22: Decision Tree Analysis

Decision Illustration

22

Sunny wants to join WMP from IIM Lucknow, Noida Campus.

She is hopeful that if after completion of course she will get better opportunity and her salary will be INR 50,00,000, if the economy is good. If the economy is average, she will get a salary of Rs. 40,00,000. If economy is bad she will get Rs. 30,00,000. The fee for course is approx. Rs. 8,10,000. Also she estimates that there would be some incidental expenses of Rs. 2,90,000 on commuting etc.In case she does not enroll for the course she will get increment on her current salary of Rs. 20,00,000 @ 30%, 20% or 10% incase of economy is good, average or bad during the duration of the course. The probability of economy to be good or bad is 30% each and to be average is 40%

Page 23: Decision Tree Analysis

23

Payoff Table

A payoff table provides alternatives,

states of nature, and payoffs

Alternative

(Action)

Salary in INR 100,000

Choice (Action)

Good

Economy

Average

Economy

Bad

Economy

Join 39 29 19

Not Join 26 24 22

Probabilities 0.3 0.4 0.3

Page 24: Decision Tree Analysis

Decision Making - Criteria

24

• Maximax

– An optimistic decision criteria

• Maximin

– A pessimistic decision criteria

• Minimax Regret

– Minimum of worst regrets

• Expected Monetary Value (EMV)

– The expected profit for taking action

• Expected Opportunity Loss (EOL)

– The expected opportunity loss for taking action.

• Expected Profit Under Certainty (EPUC)

– The expected opportunity loss from the best decision

• Expected Value of Perfect Information (EVPI)

– The expected opportunity loss from the best decision

Page 25: Decision Tree Analysis

25

Decision Tree

Decision Tree

A Decision Tree is a chronological representation of the decision process.

A Visual Representation of Alternatives, Payoffs, and Probabilities.

25

Page 26: Decision Tree Analysis

• A Decision Tree is a chronological representation of the decision process.

• The tree is composed of nodes and branches.

A branch emanating from a state of

nature (chance) node corresponds to a

particular state of nature, and includes

the probability of this state of nature.

Decision

node

Chance

nodeP(S2)

P(S2)

A branch emanating from a

decision node corresponds to a

decision alternative. It includes a

cost or benefit value.

Decision Tree

26

Page 27: Decision Tree Analysis

Decision Tree

50L

40L

25L

26L

24L

22L

29L

24L

0.3

0.4

0.3

0.3

0.4

0.3

29L11L

Join WMP

Decision Point

Action

Expected Value

27

Page 28: Decision Tree Analysis

28

Kaun Banega Crorepati

You are a contestant on “Kaun Bangega Crorepati?” You already have answered the Rs.25L question correctly and now must decide if you would like to answer the Rs. 50Lquestion. You can choose to walk away at this point with Rs. 25L in winnings or you maydecide to answer the Rs. 50L question. If you answer the Rs. 50L question correctly, youcan then choose to walk away with Rs. 50L in winnings or go on and try to answer the Rs.100L question. If you answer the Rs. 100L question correctly, the game is over and youwin Rs. 100L. If you answer either question incorrectly, the game is over immediately andyou take home “only” Rs. 3.2L.

You have the “phone a friend” lifeline remaining. With this option, you may phone afriend to obtain advice on the correct answer to a question before giving your answer. Youmay use this option only once (i.e., you can use it on either the Rs. 50L question or the Rs.100L). Since some of your friends are smarter than you are, “phone a friend” significantlyimproves your odds for answering a question correctly. Without “phone a friend,” if youchoose to answer the Rs. 50L question you have a 65% chance of answering correctly,and if you choose to answer the Rs. 100L question you have a 50% chance of answeringcorrectly (the questions get progressively more difficult). With “phone a friend,” you havean 80% chance of answering the Rs. 50L question correctly and a 65% chance ofanswering the Rs. 100L question correctly.

Page 29: Decision Tree Analysis

29

Kaun Banega Crorepati

Crt 50%

Incrt 50%

w/o Life

Crt 65%

Incrt 35%

Crt 50%

Incrt 50%

Don’t Play

100L

3.2L

50L

3.2L

100L

3.2L100L

3.2L

50L

3.2L

25L

Decision Point

Decision Point

Events

Action

Page 30: Decision Tree Analysis

30

44.10

41.92

44.10

51.60

66.12

Kaun Banega Crorepati

51.60

Crt 50%

Incrt 50%

66.12

51.60w/o Life

Crt 65%

Incrt 35%

Crt 50%

Incrt 50%

Don’t Play

100L

3.2L

50L

3.2L

100L

3.2L100L

3.2L

50L

3.2L

25L

66.12

51.6

Decision PointEvents

Action

Page 31: Decision Tree Analysis

31

References

Ernest Urbanovich, Ella E. Young, Martin L. Puterman, Sidney O. Fattedad, (2003) Early Detection of High-Risk Claims at the Workers' Compensation Board of British Columbia. Interfaces 33(4):15-26. http://dx.doi.org/10.1287/inte.33.4.15.16372

Robert K. Perdue, William J. McAllister, Peter V. King, Bruce G. Berkey, (1999) Valuation of R and D Projects Using Options Pricing and Decision Analysis Models. Interfaces 29(6):57-74. http://dx.doi.org/10.1287/inte.29.6.57

Michael R. Walls, G. Thomas Morahan, James S. Dyer, (1995) Decision Analysis of Exploration Opportunities in the Onshore US at Phillips Petroleum Company. Interfaces 25(6):39-56. http://dx.doi.org/10.1287/inte.25.6.39

Page 32: Decision Tree Analysis

Annexure