Chapter 19 Decision Analysis

LEARNING OBJECTIVES Chapter 19 describes how to use decision analysis to improve management decisions, thereby enabling you to: 1. 2. ". $. &. Learn about decision making under certainty, under uncertainty, and under risk. Learn several strategies for decision making under uncertainty, including e!pected payoff, e!pected opportunity loss, ma!imin, ma!ima!, and minima! regret. Learn how to construct and analy#e decision trees. %nderstand aspects of utility theory. Learn how to revise probabilities with sample information. C A!TER O"TLINE 19.1 'he (ecision 'able and (ecision )aking %nder Certainty (ecision 'able (ecision )aking %nder Certainty (ecision )aking %nder %ncertainty )a!ima! Criterion )a!imin Criterion *urwic# Criterion )inima! +egret (ecision )aking %nder +isk (ecision 'rees ,!pected )onetary -alue .,)-/ ,!pected -alue of 0erfect 1nformation %tility +evising 0robabilities in Light of 2ample 1nformation ,!pected -alue of 2ample 1nformation #E$ TER%S (ecision 3lternatives (ecision 3nalysis (ecision )aking %nder Certainty (ecision )aking %nder +isk (ecision )aking %nder %ncertainty (ecision 'able (ecision 'rees ,)-5er ,!pected )onetary -alue .,)-/ ,!pected -alue of 0erfect 1nformation ,!pected -alue of 2ample 1nformation *urwic# Criterion )a!ima! Criterion )a!imin Criterion )inima! +egret 4pportunity Loss 'able 0ayoffs 0ayoff 'able +isk 3voider +isk 'aker 2tates of 6ature %tility

19.2

19."

19.$

91

92

2olutions )anual and 2tudy 7uide

ST"D$ &"ESTIONS 1. 1n decision analysis, decision making scenarios are divided into three categories: decision making under 88888888888888, decision making under 888888888888888, and decision making under 88888888888888. 2. )any decision analysis problems can be viewed as having three variables: 1. 888888888888888888888, 2. 88888888888888888888, and ". 8888888888888888888888. ". 4ccurrences of nature that can happen after a decision has been made that can affect the outcome of the decision and over which the decision maker has little or no control are called 888888888888888888888888888888888. 'he benefits or rewards that result from selecting a particular decision alternative are called 888888888888888. 'he various choices or options available to the decision maker in any given problem situation are called 8888888888888888888888888888888888. $. ,!amine the decision table shown below: 2tate of 6ature 1 2 9&: 92& 1: 1& 92: 9& " ;& 2: 1: $ 12& 2& 2:

(ecision 3lternative

1 2 "

'he selected decision alternative using a )a!ima! criterion is 888888888888888888 and the optimal payoff is 88888888888. &. %se the decision table from <uestion $. 'he selected decision alternative using a )a!imin criterion is 8888888888888 and the payoff for this is 888888888888. 2uppose *urwic# criterion is used to select a decision alternative and is .". 'he selected decision alternative is 88888888888888888888 and the payoff is 88888888888888888888. *owever, if is .=, the selected decision alternative is 88888888888888888888 and the payoff is 88888888888888888888. >. %se the decision table from <uestion $ to construct an 4pportunity Loss table. %sing this table and a )inima! +egret criterion, the selected decision alternative is 888888888888888888888 and the minimum regret is 8888888888888. ;. ?ith decision trees, the decision alternatives are depicted by a 88888888888888 node and the states of nature are represented by a 8888888888888888 node. =. 'he decision table presented in <uestion $ has been reproduced below with probabilities assigned to the states of nature: 2tate of 6ature 1..2:/ 2.."&/ "..$:/ $..:&/ 9&: 92& ;& 12& 1: 1& 2: 2& 92: 9& 1: 2:

(ecision 3lternative

1 2 "

Chapter 19: (ecision 3nalysis

9"

'he e!pected monetary value of selecting decision alternative 1 is 888888888888888. 'he e!pected monetary value of selecting decision alternative 2 is 888888888888888. 'he e!pected monetary value of selecting decision alternative " is 888888888888888. 9. 3n ,)-5er would choose decision alternative 8888888888 based on the results of <uestion =. 1:. 'he e!pected value of perfect information for the decision scenario presented in <uestion = is 8888888888888888888. 11. 'he degree of pleasure of displeasure a decision maker has in being involved in the outcome selection process given the risks and opportunities available is 888888888888888888. 12. 1f it takes more than the e!pected monetary value to get a player to withdraw from a game, then that person is said to be a 8888888888888888. 1". Consider the decision table shown below: 2tate of 6ature @..$:/ A..>:/ 2:: &:: &: =::

(ecision

1 2

'he e!pected monetary value of this decision scenario is 8888888888888888888. 2uppose the decision maker has an opportunity to buy a forecast to assist himBher in making the decision. ?hen the state of nature is @, the forecaster will predict @ =&C of the time and A 1&C of the time. ?hen the state of nature is A, the forecaster will predict A 9&C of the time and will predict @ &C of the time. 1f the forecast is purchased and the forecaster predicts that @ will occur, then revised probability of @ occurring is 8888888888888888 and of A occurring is 8888888888888. 1f the forecaster predicts that A will occur, then the revised probability of @ occurring is 8888888888888 and of A occurring is 888888888888888888. 'he probability that the forecaster will predict @ is 888888888888888. 'he probability that the forecaster will predict A is 88888888888888. 'he e!pected monetary value with information is 88888888888888888. 'he e!pected value of sample information is 88888888888888888.

9$

2olutions )anual and 2tudy 7uide

ANS'ERS TO ST"D$ &"ESTIONS 1. 2. ". $. &. >. Certainty, %ncertainty, +isk (ecision 3lternatives, 2tates of 6ature, 0ayoffs 2tates of 6ature, 0ayoffs, (ecision 3lternatives 1, 12& 2, 1:, 2, 1$.&, 1, 9: 1, >: ;. =. 9. 1:. 11. 12. 1". , ! 1;.&, 1>.&, 9:.;& 1 2> %tility +isk taker &::, .91=9, .:=11, .:9&2, .9:$=, .";, .>", .&$2.:2, $2.:2

SOL"TIONS TO ODD(N"%BERED !ROBLE%S IN C A!TER 19 19.1 s1 d1 2&: d2 11: d" "9: s2 1;& 1:: 1$: s" )a! )in 92& 2&: 92& ;: 11: ;: 9=: "9: 9=: decision: 2elect d" decision: 2elect d2

a./ )a! D2&:, 11:, "9:E F )9* b./ )a! D92&, ;:, 9=:E F +* c./ Gor F ." d1: .".2&:/ H .;.92&/ F &;.& d2: .".11:/ H .;.;:/ F ,d": ."."9:/ H .;.9=:/ F >1 decision: 2elect d2 Gor F .= d1: .=.2&:/ H .2.92&/ F 19& d2: .=.11:/ H .2.;:/ F 1:2 d": .=."9:/ H .2.9=:/ F -9. decision: 2elect d"

Comparing the results for the two different values of alpha, with a more pessimist point of view . F ."/, the decision is to select d2 and the payoff is =2. 2electing by using a more optimistic point of view . F .=/ results in choosing d" with a higher payoff of 29>.

Chapter 19: (ecision 3nalysis

9&

d./ 'he opportunity loss table is: s1 d1 1$: d2 2=: d" : s2 : ;& "& s" 9& : 1&: )a! 1$: 2=: 1&:

'he minima! regret F min D1$:, 2=:, 1&:E F 1/* (ecision: 2elect d1 to minimi#e the regret. 19." 3 I C ( + >: 1: 91: 2: ( 1& 2& $: 2& 1 92& ": 1& & )a! >: ": $: 2& )in 92& 1: 91: &

)a!ima! F )a! D>:, ":, $:, 2&E F .* (ecision: 2elect 3 )a!imin F )a! D92&, 1:, 91:, &E F 1* (ecision: 2elect I 19.&, 19.>

9>

2olutions )anual and 2tudy 7uide

19.;

,!pected 0ayoff with 0erfect 1nformation F &..1&/ H &:..2&/ H 2:..":/ H =..1:/ H >..2:/ F )10+1 ,!pected -alue of 0erfect 1nformation F "1.2& 9 2&.2& F .01*

19.9 Lock 1n 6o

(own..":/ 91&: 1;&

%p..>&/ 2:: 92&:

6o Change..:&/ : :

,)=& 911:

(ecision: Iased on the highest ,)-/2,13, JLock 1nJ ,!pected 0ayoff with 0erfect 1nformation F 1;&..":/ H 2::..>&/ H :..:&/ F 1,-01 ,!pected -alue of 0erfect 1nformation F 1=2.& 9 =& F 9+01 19.11 a./ ,)- F 2::,:::..&/ H .9&:,:::/..&/ F +14*** b./ +isk 3voider because the ,)- is more than the investment .;&,::: K &:,:::/ c./ Aou would have to offer more than ;&,::: which is the e!pected value. 19.1" 2 ) L (ec..>:/ 922& 12& "&: 1nc..$:/ $2& 91&: 9$:: ,)"& 1& &:

(ecision: Iased on ,)- F )a!imum D"&, 1&, &:E F 1* Gor Gorecast .(ecrease/: 0rior (ecrease .>: 1ncrease .$: Cond. Loint .;& .$& .1& .:> G.(ec/ F .&1 +evised .==2$ .11;>

Gor Gorecast .1ncrease/: 0rior (ecrease .>: 1ncrease .$: Cond. Loint .2& .1& .=& ."$ G.1nc/ F .$9 +evised .":>1 .>9"9

Chapter 19: (ecision 3nalysis

9;

'he e!pected value with sampling is -//0-+1 ,-21 F ,-?2 9 ,)- F 2$$.2;& 9 &: F 19/0-+1 19.1& (rill (on5t (rill 4il..11/ 1,:::,::: : 6o 4il..=9/ 91::,::: : ,)21,::: :

(ecision: 'he ,)- for this problem is )a! D21,:::, :E F 21,:::. 'he decision is to (rill.

9=

2olutions )anual and 2tudy 7uide

4il Gorecast 6o 4il Gorecast 4il: 2tate 4il 6o 4il 0rior .11 .=9

3ctual 4il 6o 4il .2: .1: .=: .9:

Cond. Loint .2: .:22 .1: .:=9 0.G4il/ F .111

+evised .19=2 .=:1=

Gorecast 6o 4il: 2tate 4il 6o 4il 0rior .11 .=9 Cond. Loint .=: .:== .9: .=:1 0.G6o 4il/ F .==9 +evised .:99: .9:1:

'he ,!pected -alue ?ith 2ampling 1nformation is -14*1-0),-21 F ,-?21 9 ,)- F 21,::: 9 21,:12."2 F 1-0)-

Chapter 19: (ecision 3nalysis

99

19.1;

b./ d1: $::.."/ H 2&:..2&/ H "::..2/ H 1::..2&/ F -.+01 d2: "::.."/ H .91::/..2&/ H >::..2/ H 2::..2&/ F 2"& (ecision: 2elect d1 c./ ,!pected 0ayoff of 0erfect 1nformation: $::.."/ H 2&:..2&/ H >::..2/ H 2::..2&/ F )1-01 -alue of 0erfect 1nformation F "&2.& 9 2>;.& F ,1 19.19 2mall )odest Large 2mall 2:: 1:: 9":: )oderate 2&: ":: $:: Large ":: >:: 2::: )in 2:: 1:: 9":: )a! ":: >:: 2:::

a./ )a!ima!: )a! D"::, >::, 2:::E F -*** (ecision: Large 6umber )inima!: )a! D2::, 1::, 9"::E F -** (ecision: 2mall 6umber

1::

2olutions )anual and 2tudy 7uide

b./ 4pportunity Loss: 2mall )odest Large 2mall : 1:: &:: )oderate 1&: 1:: : Large 1;:: 1$:: : )a! 1;:: 1$:: &::

)in D1;::, 1$::, &::E F 1** (ecision: Large 6umber c./ )inima! regret criteria leads to the same decision as )a!ima!. 19.21 +eg. ?eekend 6ot 4pen )ild..;&/ 2::: 12:: 9":: 2evere..2&/ 92&:: 92:: 1:: ,)=;& =&: 92::

(ecision: Iased on )a! ,)- F )a!D=;&, =&:, 92::E F ,+1, open regular hours.

,!pected -alue with 0erfect 1nformation F 2:::..;&/ H 1::..2&/ F 11-1 -alue of 0erfect 1nformation F 1&2& 9 =;& F .1*

Chapter 19: (ecision 3nalysis

1:1

19.2" 3utomate (o 6ot

+ed...1&/ 9$:,::: &,:::

Con..."&/ 91&,::: 1:,:::

1nc...&:/ >:,::: 9":,:::

,)1=,;&: 91:,;&:

(ecision: Iased on )a! ,)- F )a! D1=;&:, 91:;&:E F 1,4+1*, 2elect 3utomate Gorecast +eduction: 2tate + C 1 0rior .1& ."& .&: Cond. Loint .>: .:9 .1: .:"& .:& .:2& 0.G+ed/ F .1&: +evised .>: .2""" .1>>;

Gorecast Constant: 2tate + C 1 0rior .1& ."& .&: Cond. Loint .": .:$& .=: .2=: .2& .12& 0.GCons/ F .$&: +evised .1: .>222 .2;;=

Gorecast 1ncrease: 2tate + C 1 0rior .1& ."& .&: Cond. Loint +evised .:";& .:=;& .=;&:

.1: .:1& .1: .:"& .;: ."&: 0.G1nc/ F .$::

1:2

2olutions )anual and 2tudy 7uide

,!pected -alue ?ith 2ample 1nformation F -14/-1011 ,-21 F ,-?21 9 ,)- F 21,$2&.&& 9 1=,;&: F -4.+1011

Chapter 19: (ecision 3nalysis

1:"

1:$

2olutions )anual and 2tudy 7uide

Chapter 19: (ecision 3nalysis

1:&

1:>

2olutions )anual and 2tudy 7uide

Chapter 19: (ecision 3nalysis

1:;

1:=

2olutions )anual and 2tudy 7uide