Anda di halaman 1dari 74

Six Steps of Decision Making

Clearly define the problem at hand List the possible alternatives Identify the possible outcomes List the payoff or profit of each combination of alternatives and outcomes Select one of the mathematical decision theory models Apply the model and make your decision

Relationships
Uncertainty
The Environment Goals and Objectives

Future Conditions
Courses of Action Result

Preferences
Value

Alternatives

Outcomes

Payoff

Structuring Decisions Probability Decisions Under Uncertainty Models of Uncertainty

Relationships
Tradeoffs
The Environment Goals and Objectives

Future Conditions
Courses of Action Result

Preferences
Value

Alternatives

Outcomes

Payoff

Multi-Objective Decision Making Quantifying Preferences Effectiveness Analysis Cost-Effectiveness Analysis

Terms used to describe the uncertainty:


undetermination - where estimations can be absolutely certain, but in a probabilistic sense. Rationality does not fully guarantee absolute success, it merely indicates a higher probability of success; ambiguity - refers to the decision-makers inability to clearly determine the significance of the situations he will have to deal with; risk - represents the more or less conscious assuming of the outcomes of the choice made; uncertainty - can represent the limited explicative-predictive capacity of the decision-makers knowledge, be this limitation originating either in the ontological indetermination or in the approximate character of the decision-makers knowledge at a given moment.

From the point of view of human behavior in the decision-making process


Objective uncertainty can be considered a measure of the real knowledge the decisionmaker possesses; it is expressed by a ratio between the quantity and the quality of the information required to make decisions and the information the decision-maker actually has; it cannot be expressed precisely only in relation with a theoretical situation. Subjective uncertainty refers to how aware the decision-maker is of his degree of certainty/uncertainty; the decision-makers ability to express himself is determined by his professional competence, his economic and general/background knowledge, his field-related experience, his creativity, his ability to analyze and synthesize, his inquisitive and normative line of thought.

Classifying Decisions, there are:


Decisions under certainty (all the facts are known for sure) uncertainty (the possible events are not
known for sure but probabilities can be assigned to the possible occurrences)

Classifying Decisions, there are:


Decisions made for - one time period only or - a sequence of interrelated decisions over several time periods (called a dynamic decision problem Decisions where the opponent is - nature (drilling an oil well) or - a thinking opponent (setting a budget and counting on what the competitors are to develop)

TYPES OF DECISION - MAKING ENVIRONMENTS Decision - making under certainty Decision - making under risk

Decision - making under uncertainty

Decisions and Probability


The quantitative analysis which supports decision making under certainty usually takes the form of maximizing some objective (say profit or production) subject to constraints (productive capacity) When the true state of nature is unknown the decision maker has to act with imperfect information

Relationships: Decision matrix


Future Condition 1 Alternative 1 Alternative 2 Future Condition 2
Probability Outcome (Payoff) Probability Outcome (Payoff)

...

Future Condition N

Alternative K

...
Probability Outcome (Payoff)

... ...

ROMANIAN AERONAUTICAL AUTHORITY (R.A.A.)


MUST CHOOSE FROM AMONG A NUMBER OF BOMB DETECTION SYSTEMS FOR COMMERCIAL AIR CARRIERS

R.A.A. MUST TRY TO ACCOMPLISH NEXT OBJECTIVES: 1. THE CHOSEN SYSTEM HAS TO BE EFFECTIVE AS POSSIBLE AT DETECTING VARIOUS TYPES OF EXPLOSIVES 2. TO IMPLEMENT THE SYSTEM AS QUICKLY AE POSSIBLE 3. TO MAXIMIZE PASSENGER ACCEPTANCE 4. TO MINIMIZE THE COST

Why use a tree?


Safe job Pass Test pPass | Study Fail Test Safe job Pass Test pPass | Dont Fail Test Highly visible job Succeed pSucceed Fail Highly visible job Succeed pSucceed Fail Study for Test

Don't Study For Test

Certainty?
Fixed determined 101% Sure of Unquestionable Reliable dependable Without doubt Controlled Will Happen

Difficult Questions
(Any Decision Maker Will Do) What scenarios should we plan for? What approach should be taken to deterrence? Are there challenges to our peacemaking? How low can our defense budget afford to go? Do we need a large ground/air/ sea force? What are threats to our sovereignty (now-later)? Should we re-engineer our defense dept? What should be our active / reserve force mix?

DECISIONS UNDER CERTAINTY


ADVANTAGES it is conceptually simple; has made possible to handle a problem with a large number of restrains in an orderly way; has exceptional power and generality; can be solved using computers. LIMITATIONS the solutions are not always integer values; the uncertainty is not allowed; the assumption of linearity. CONCLUSIONS linear programming cannot be applied to all problems, but is a powerful tool for those problems where it is applicable.

Decision under Certainty Case Study Two types of aircraft are available for a bombing mission. Bombs are plentiful but aircraft, fuel and maintenance capacity are limited. Only 4000 gallons of fuel are available and no more than 30 planes can receive the required maintenance. Aircraft type A B # avaliable 26 15 Payload 2 tons 3 tons Fuel consumption 100 gallons 200 gallons

Steps to deal with uncertainty


Step Identify uncertainty Measure uncertainty Model process Decide Lecture Structuring Decisions Probability Models of Uncertainty Decisions Under Uncertainty

DECISION MAKING UNDER UNCERTAINTY Criteria: n Maximax o Maximin p Equally likely q Minimax

The Central Challenge? Planning Under Uncertainty!

Maximax decision rules


Optimistic assessment: assumes favorable future conditions decision maker selects BEST OF THE BEST maximizes potential gain may expose decision maker to great risk

Maximin decision rules


Pessimistic assessment: assumes unfavorable future conditions decision maker selects BEST OF THE WORST appropriate when decision maker involves possibility of catastrophic outcomes involves strong aversion to risk

Hurwicz (combination) decision rules


Combine optimism with pessimism: assumes for both favorable / unfavorable future conditions a degree of: optimism ( - expressed as fraction) pessimism (1-)

Minimax Regret decision rules


Conservative assessment: a strategy for risk minimization: minimize maximum possible after the fact regret associated with a wrong decision focuses on opportunity cost requites creation of a regret matrix decision maker selects the alternative that minimize maximum regret

Laplace decision rules


Neither optimistic or pessimistic: assumes that all future conditions have equal likelihood of occurring (insufficient reason criterion) create an average (combined) payoff measure and decision maker selects the alternative with best combined average

Maximum likelihood decision rules


we know the likelihood of occurrence of each future condition (consider payoff and probability) identify the future condition with the highest probability of occurrence select the alternative with the highest payoff for that future condition for short run, one time decisions

Expected value of alternative


we know the likelihood of occurrence of each future condition multiply the probability by the payoff of each future condition and then add them together select the alternative with the highest score with no clear preference, long run, repetitive decisions

Decision under uncertainty


- uncertainty is not bad, is real - it can be managed, not eliminated - we must understand and accept the risk If we (dont) know the likelihood of occurrence of each future condition

Six Steps of Decision Making


Clearly define the problem at hand List the possible alternatives Identify the possible outcomes List the payoff or profit of each combination of alternatives and outcomes Select one of the mathematical decision theory models Apply the model and make your decision

Elements of a decision problem


Goals and objectives Alternatives; courses of actions Future conditions; states of nature; uncertainty Outcomes; consequences Payoffs; gain and losses; measures of value Values and preferences

Classifying Decisions, there are:


Decisions under certainty (all the facts are known for sure) uncertainty (the possible events are not known for sure but probabilities can be assigned to the possible occurrences)

Decision Matrix Under Certainty

Future Condition Alternative 1 Alternative 2 Outcome/payoff? Outcome/payoff?

Alternative K

...
Outcome/payoff?

Decision under uncertainty


- uncertainty is not bad, is real - it can be managed, not eliminated - we must understand and accept the risk If we (dont) know the likelihood of occurrence of each future condition

Maximax decision rules


Optimistic assessment: assumes favorable future conditions decision maker selects BEST OF THE BEST maximizes potential gain may expose decision maker to great risk

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

13 15 25 24

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

13 15 25 24

Maximin decision rules


Pessimistic assessment: assumes unfavorable future conditions decision maker selects BEST OF THE WORST appropriate when decision maker involves possibility of catastrophic outcomes involves strong aversion to risk

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

13 15 25 24

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

13 15 25 24

5 3 1 4

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

13 15 25 24

5 3 1 4

Hurwicz (combination) decision rules


Combine optimism with pessimism: assumes for both favorable / unfavorable future conditions a degree of: optimism ( - expressed as fraction) pessimism (1-) =70% or =0.7 1-=0.3

Decision matrix
F1 F2 F3 F4 MAXI MAXI HURWI MAX MIN CZ MINI LAPL MAX ACE REG RET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

13 15 25 24

5 3 1 4

13*0.7 +5*0.3

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

13 15 25 24

5 3 1 4

10.6 11.4 17.8 18

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

13 15

5
3 1 4

10.6 11.4 17.8 18

25
24

Minimax Regret decision rules


Conservative assessment: a strategy for risk minimization: minimize maximum possible after the fact regret associated with a wrong decision focuses on opportunity cost requites creation of a regret matrix decision maker selects the alternative that minimize maximum regret

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 (3) 9 (0) 4 (5) 5 (4)

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 (3) 9 (0) 4 (5) 5 (4)

10 (2) 12 (0) 2 4 (8)

13 3 1 24 (0)

5 15 25 5 (20)

20 21 23 20

(11) (20) (21) (10)

(10) (23) (0)

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 (3) 9 (0) 4 (5) 5 (4)

10 (2) 12 (0) 2 4 (8)

13 3 1 24 (0)

5 15 25 5 (20)

(11) (20) (21) (10)

20 11 21 23 20 8

(10) (23) (0)

Laplace decision rules


Neither optimistic or pessimistic: assumes that all future conditions have equal likelihood of occurring (insufficient reason criterion) create an average (combined) payoff measure and decision maker selects the alternative with best combined average

Decision matrix
F1 F2 F3 F4 MAX MAX LAPLACE IMA IMIN X

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

(6+10+13+5)/4 =34/4

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

34/4 39/4 32/4 38/4

Decision matrix
F1 F2 F3 F4 MAXI MAXI HUR MINI LAPL MAX MIN WICZ MAX ACE REGR ET

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

34/4 39/4 32/4 38/4

Maximum likelihood decision rules


we know the likelihood of occurrence of each future condition (consider payoff and probability) identify the future condition with the highest probability of occurrence select the alternative with the highest payoff for that future condition for short run, one time decisions

Decision matrix
F1 90% F2 3% F3 4% F4 4% MAXIMUM LIKLIHOOD

A1 A2 A3 A4

6 () 9 () 4 () 5 ()

10 () 12 () 2 () 4 ()

13 () 3 () 1 () 24 ()

5 () 15 () 25 () 5 ()

6 9 4 5

Decision matrix
F1 0.2 F2 0.3 F3 0.2 F4 0.3 MAXIMUM LIKLIHOOD

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

Decision matrix
F1 0.2 F2 0.3 F3 0.2 F4 0.3 MAXIMUM LIKLIHOOD

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

10 15 25 5

Expected value of alternative


we know the likelihood of occurrence of each future condition multiply the probability by the payoff of each future condition and then add them together select the alternative with the highest score with no clear preference, long run, repetitive decisions

Decision matrix
F1 0,2 F2 0,1 F3 0,6 F4 0,1 Expectrd value

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

6*0.2+10*0.1+1 3*0.6+5*0.1

Decision matrix
F1 0,2 F2 0,1 F3 0,6 F4 0,1 Expectrd value

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

10.5 6.3 4.1 16.3

Decision matrix
F1 0,2 F2 0,3 F3 0,2 F4 0,3 Expectrd value

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

Decision matrix
F1 0,2 F2 0,3 F3 0,2 F4 0,3 Expectrd value

A1 A2 A3 A4

6 9 4 5

10 12 2 4

13 3 1 24

5 15 25 5

8.3 10.5 9.1 8.5

Case study
The Army plans to procure Armored Personnel Carriers (APCs) from one of five foreign suppliers. Fluctuations in the country exchange rate relative to the value of the source countrys currency could dramatically alter both the expected procurement cost and future years operations and support cost of these APCs. The costs (payoffs) shown below represent the estimated change (in millions of dollars) to the armys currently proposed procurement and O&S budget based on possible fluctuations in the value of their currency on the world market.

The Army plans to procure Armored Personnel Carriers (APCs) from one of five foreign suppliers. Fluctuations in the country exchange rate relative to the value of the source countrys currency could dramatically alter both the expected procurement cost and future years operations and support cost of these APCs. The costs (payoffs) shown below represent the estimated change (in millions of dollars) to the armys currently proposed procurement and O&S budget based on possible fluctuations in the value of their currency on the world market.

Country 1 Country 2 Country 3 Country 4 Country 5

Major Minor Some Major appreciatio appreciatio depreciatio depreciatio n n n n -5 2 8 10 -15 1 0 -1 4 2 4 6 14 13 16 17 27 20 21 22

Case study
Which alternative source country would you choose if: you followed a mini-min strategy ? you followed a mini-max strategy ?

Case study
Major Minor Some Major appreciatio appreciatio depreciatio depreciatio n n n n B-5 B+2 B+8 B+10 B-15 B+1 B0 B-1 B+4 B+2 B+4 B+6 B+14 B+13 B+16 B+17 B+27 B+20 B+21 B+22

Country 1 Country 2 Country 3 Country 4 Country 5

Case study
Major Minor Some Major app. app. dep. dep.

MINI MINI Hur Mini MIN MAX wicz Regre t

C1 C2 C3 C4 C5

B-5

B+2

B+8 B+1 4 B+1 3 B+1 6 B+1 7

B-15 B+4 B+1 B0 B-1 B+2 B+4 B+6

B+1 0 B+2 7 B+2 0 B+2 1 B+2 2

Case study
Major Minor Some Major app. app. dep. dep.

C1 C2 C3 C4 C5

B-5 (5) B-15 (15) B+1 (-1) B0 (0) B-1 (1)

B+2 (0) B+4 (2) B+2 (0) B+4 (2) B+6 (4)

B+8 (0) B+1 4 (6) B+ 1 3 (5) B+ 1 6 (8) B+ 1 7 (9)

MINI MINI Hur Mini MIN MAX wicz Max Regre t B+1 2 0 (0) B+ 2 4 7 (17) B+ 2 1 0 (10) B+ 2 0 1 (11) B+ 2 -1 2 (12)

Case study
Which alternative source country would you choose if: you followed a mini-max strategy ? you followed a mini-min strategy ?

Case study
Major Minor Some Major app. app. dep. dep.

C1 C2 C3 C4 C5

B-5 (5) B-15 (15) B+1 (-1) B0 (0) B-1 (1)

B+2 (0) B+4 (2) B+2 (0) B+4 (2) B+6 (4)

B+8 (0) B+1 4 (6) B+ 1 3 (5) B+ 1 6 (8) B+ 1 7 (9)

MINI MINI Hur Mini MIN MAX wicz Max Regre t B+1 2 10 0 (0) B+ 2 4 27 7 (17) B+ 2 1 20 0 (10) B+ 2 0 21 1 (11) B+ 2 -1 22 2 (12)

Case study
Which alternative source country would you choose if: you followed a mini-max strategy ? you followed a mini-min strategy ? you followed a strategy with a coefficient of optimism of 70 % ?

Case study
Major Minor Some Major app. app. dep. dep.

C1 C2 C3 C4 C5

B-5 (5) B-15 (15) B+1 (-1) B0 (0) B-1 (1)

B+2 (0) B+4 (2) B+2 (0) B+4 (2) B+6 (4)

B+8 (0) B+1 4 (6) B+ 1 3 (5) B+ 1 6 (8) B+ 1 7 (9)

MINI MINI Hur Mini MIN MAX wicz Max Regre B+1 2 10 0.7*2 t +0.3* 0 10=4. (0) 4 B+2 4 27 10.9 7 (17) B+ 2 1 20 6.7 0 (10) B+ 2 0 21 6.3 1 (11) B+ 2 -1 22 7.3 2 (12)

Case study
Which alternative source country would you choose if: you followed a mini-max strategy ? you followed a mini-min strategy ? you followed a strategy with a coefficient of optimism of 70 % ? you minimize your regret ?

Case study
Major Minor Some Major app. app. dep. dep.

C1 C2 C3 C4 C5

B-5 (5) B-15 (15) B+1 (-1) B0 (0) B-1 (1)

B+2 (0) B+4 (2) B+2 (0) B+4 (2) B+6 (4)

B+8 (0) B+1 4 (6) B+ 1 3 (5) B+ 1 6 (8) B+ 1 7 (9)

MINI MINI Hur Mini MIN MAX wicz Max Regre 2 10 0.7*2 t 5 B+1 +0.3* 0 10=4. (0) 4 4 27 10.9 17 B+2 7 (17) 1 20 6.7 10 B+ 2 0 (10) 0 21 6.3 11 B+ 2 1 (11) B+ 2 -1 22 7.3 12 2 (12)

???
You assumed that each future condition was equally likely ? What was your objective function ?

Anda mungkin juga menyukai