Decision Tree

Penelitian Operasional 3 (ISE – 183101)

Reference:
Winston, W.L. & Goldberg, J.B. (2004). Operations research: applications and algorithms (Vol 4).
BelmonteCalif Calif:Thomson/Brooks/Cole.

Chapter 13.4
Your Logo or Name Here
Outline

Utility and Risk Preferences

Decision Tree

Incorporating Risk Aversion into Decision Tree Analysis

Expected Value of Sample Information (EVSI) and

Expected Value of Perfect Information (EVPI)

Your Logo or Name Here

2
Utility Function and Risk Preferences
• Which one do you prefer? ❑ Utility u(rn)→ measures individual
preferences over a reward/ payoff.
100%
C $ 200.000 ❑ Utility represents the satisfaction that
individual receives for choosing and getting
or a certain reward/payoff.
50%
$ 100.000 ❑ The higher utility, the higher preference is.
L ❑ Risk Preferences:
50%
$ 300.000 ❑ Risk adverse
❑ Risk neutral/ indifference
EV (L) = 0.5*100,000 + 0.5*300,000
= $ 200,000 ❑ Risk lover/ seeker
Your Logo or Name Here
3
Utility Function and Risk Preferences – Risk Averse
Utility u(x) with certainty ❑ Risk averse: decrease in marginal
u(x) of EV of lotery utility.

❑ Certainty equivalent of a lottery L,

CE(L), is the number CE(L) such that the
decision maker is indifferent between
the lottery L and receiving a certain
payoff of CE(L).

❑ The risk premium of a lottery L,

RP(L) = EV(L) - CE(L)

❑ RP (L) = 200,000 – 145,000 =55,000.

Reward
❑ Risk averse → RP (L) > 0.
$100,000 $200,000 $300,000
$145,000 → CE (L) Your Logo or Name Here
4
Utility Function and Risk Preferences – Risk Neutral
Utility ❑ Risk Neutral: constant marginal utility.
u(x) with certainty = u(x) of EV of lotery

❑ Risk Neutral → RP (L) = 0.

Reward

$100,000 $200,000 $300,000

Your Logo or Name Here
5
Utility Function and Risk Preferences – Risk Lover
Utility u(x) with certainty ❑ Risk lover: Increasing in marginal
u(x) of EV of lotery utility.

❑ Certainty equivalent of a lottery L,

CE(L), is the number CE(L) such that the
decision maker is indifferent between
the lottery L and receiving a certain
payoff of CE(L).

❑ The risk premium of a lottery L,

RP(L) = EV(L) - CE(L)

❑ RP (L) = 200,000 – 231,920 = - 31,920.

❑ Risk lover → RP (L) < 0.

Reward
❑ Utility function, u(x), is strictly convex.
$100,000 $200,000 $300,000
$231,920 → CE (L) Your Logo or Name Here
6
Utility Function and Risk Preferences (5)
Utility

❑ Risk averse: Utility function, u(x), is

strictly concave.

❑ Risk Neutral :Utility function, u(x), is

linear function.

❑ Risk lover: Utility function, u(x), is

strictly convex.

Reward

$100,000 $200,000 $300,000

Your Logo or Name Here
7
Outline

Utility and Risk Preferences

Decision Tree

Incorporating Risk Aversion into Decision Tree Analysis

Expected Value of Sample Information (EVSI) and

Expected Value of Perfect Information (EVPI)

Your Logo or Name Here

8
Decision Tree
❑ Used to determine optimal decisions.

❑ A decision tree enables a decision maker to decompose a large complex decision problem
into several smaller problems.

Your Logo or Name Here

9
Example Colaco Marketing

Colaco currently has assets of $150,000 and wants to decide whether to market a
new chocolate-flavored soda, Chocola. Colaco has three alternatives:
❑ Alternative 1 Test market Chocola locally, then utilize the results of the
market study to determine whether or not to market Chocola nationally.
❑ Alternative 2 Immediately (without test marketing) market Chocola
nationally.
❑ Alternative 3 Immediately (without test marketing) decide not to market
Chocola nationally.

Your Logo or Name Here

10
Example Colaco Marketing (2)
❑ In the absence of a market study, Colaco believes that Chocola has a 55% chance of
being a national success and a 45% chance of being a national failure. If Chocola is a
national success, Colaco’s asset position will increase by $300,000, and if Chocola is
a national failure, Colaco’s asset position will decrease by $100,000.
❑ If Colaco performs a market study (at a cost of $30,000), there is a 60% chance that
the study will yield favorable results (referred to as a local success) and a 40%
chance that the study will yield unfavorable results (referred to as a local failure). If a
local success is observed, there is an 85% chance that Chocola will be a national
success. If a local failure is observed, there is only a 10% chance that Chocola will be
a national success.
❑ If Colaco is risk-neutral (wants to maximize its expected final asset position), what
strategy should the company follow?

Your Logo or Name Here

11
Drawing Decision Tree
❑ Begin at the present and proceed toward future events and
decisions.
❑ Constructed with two kinds of forks:
❑ Decision forks - denoted by
A point in time when decision maker has to make a decision. Each
branch emanating from a decision fork represents a possible
decision.

❑ Event forks - denoted by

drawn when outside forces determine which of several random
events will occur. Each branch of an event fork represents a
possible outcome, and the number on each branch represents the
probability that the event will occur.
❑ A terminal branch is if no forks emanate from the branch.
On each terminal branch, calculate the reward/payoff (based on the
objective).
Your Logo or Name Here
12
❑ Decision Tree for Colaco Marketing (Risk-Neutral)

Your Logo or Name Here

13
Evaluate the Decision Tree

❑ Working backward from right to left (“folding back the tree”) .

❑ Steps:
❑ At each event fork, we calculate the expected reward/payoff and enter it in
❑ At each decision fork, we denote by ‖ the decision that maximizes the
reward/payoff and enter the expected reward/payoff associated with that
decision in .
❑ Continue working backward in this fashion until we reach the beginning of
the tree.
❑ The optimal sequence of decisions can be obtained by following the ‖ .

Your Logo or Name Here

14
At each event fork, we
calculate the expected
reward/payoff and enter
it in
ER = 0.85*(420,000) + 0.15*(20,000) = $360,000.

ER = 0.1*(420,000) + 0.9*(20,000) = $60,000.

ER = 0.55*(450,000) + 0.45*(50,000) = $270,000.

Your Logo or Name Here

15
At each decision fork, the
decision that maximizes
the reward denoted by ‖ .
Enter the ER associated
‖
with that decision in .
ER = 0.6*(360,000) + 0.4*(120,000) = $260,000.

‖
‖

Colaco’s optimal decision is Don’t test market

Your Logo or Name Here
and then Market nationally. 16
Outline

Utility and Risk Preferences

Decision Tree

Incorporating Risk Aversion into Decision Tree Analysis

Expected Value of Sample Information (EVSI) and

Expected Value of Perfect Information (EVPI)

Your Logo or Name Here

17
Incorporating Risk Aversion into Decision Tree Analysis
• Colaco’s Decision Tree ❑ Colaco’s optimal strategy yields a 0.45
chance that the company will end up
with a relatively small final asset
position of $50,000.

❑ On the other hand, the strategy of test

marketing and acting optimally on the
results of the test market study yields
only a 0.6*0.15 = 0.09 chance that
Colaco’s asset position will be below
$100,000.

❑ Thus,if Colaco is a risk-averse

decision maker, the strategy of
immediately marketing nationally
may not reflect the company’s
preference.

Your Logo or Name Here

18
Incorporating Risk Aversion into Decision Tree Analysis
(2)
Colaco’s utility function ❑ Colaco has the risk-averse utility function.

❑ To determine optimal decisions (that is, the

decisions that maximize expected utility):
1. Replace each final asset position x0 with its
utility u(x0).
2. At each event fork, compute the expected
utility and at each decision fork, choose the
branch having the largest expected utility.

Your Logo or Name Here

19
❑ Decision Tree for Colaco Marketing (Risk-Averse)
1. Replace each final asset position
x0 with its utility u(x0).

2. At each event fork,

compute the ‖
expected utility and
at each decision fork,
choose the branch
having the largest
expected utility. ‖
‖

Your Logo or Name Here

20
❑ Decision Tree for Colaco Marketing (Risk-Averse)
1. Replace each final asset position
x0 with its utility u(x0).

2. At each event fork,

compute the ‖
expected utility and
at each decision fork,
choose the branch
having the largest
expected utility. ‖
‖

Your Logo or Name Here

21
❑ Decision Tree for Colaco Marketing (Risk-Averse)
Colaco’s optimal decision is to
begin by test marketing.

If a local success →
market nationally; ‖
if a local → should not market
nationally.

‖
‖

Your Logo or Name Here

22
Incorporating Risk Aversion into Decision Tree Analysis
(5)
Colaco’s utility function ❑ Colaco has the risk-averse utility function.

❑ From graphic → u($226,000) = 0.665.

❑ Colaco views the current situation as having an

expected utility of .6649, this means that the company
considers the current situation equivalent to a certain
asset position of $226,000.

❑ Thus, if somebody offered to pay more than $76,000

(226,000 -150,000) to buy the rights to Chocola, Colaco

should take the offer.

This is because receiving more than $76,000 for the rights

to Chocola would bring Colaco’s asset position to more
than 150,000+76,000=$226,000, and this situation has a
higher expected utility than .665. Your Logo or Name Here
23
Outline

Utility and Risk Preferences

Decision Tree

Incorporating Risk Aversion into Decision Tree Analysis

Expected Value of Sample Information (EVSI) and

Expected Value of Perfect Information (EVPI)

Your Logo or Name Here

24
Expected Value of Sample Information (EVSI)

❑ Decision trees can be used to measure the value of sample or test market
information.
❑ Steps:
1. Determine expected value with sample information (EVWSI) →
Expected reward/ payoff if the company acts optimally and the test
market study is costless.
2. Determine expected value with original information (EVWOI) → the
largest expected reward if the test market study were not available.
3. Calculate expected value of sample information (EVSI) → expected
value of the test market information.
EVSI = EVWSI - EVWOI.
Your Logo or Name Here
25
Expected Value of Sample Information (EVSI)

❑ Decision trees can be used to measure the value of sample or test market
information.
❑ Steps:
1. Determine expected value with sample information (EVWSI) →
Expected reward/ payoff if the company acts optimally and the test
market study is costless.
2. Determine expected value with original information (EVWOI) → the
largest expected reward if the test market study were not available.
3. Calculate expected value of sample information (EVSI) → expected
value of the test market information.
EVSI = EVWSI - EVWOI.
Your Logo or Name Here
26
Decision Tree for Colaco Marketing (Risk-Neutral)
1. Expected value with sample
information (EVWSI) =
264,000 + 30,000 = 294,000

2. Expected value with original

information (EVWOI)
270,000

3. Expected value of sample

information (EVSI))
294,000 - 270,000 = 24,000

Since the cost of the test market

study ($30,000) exceeds EVSI,
Colaco should not conduct the
test market study.
Your Logo or Name Here
27
Expected Value of Perfect Information (EVPI)

❑ Perfect information → all uncertain events that can affect reward still occur
with the given probabilities, but Decision maker finds out event is happening
before making the decision.
❑ As for Colaco example: all uncertain events that can affect Colaco’s final
asset position still occur with the given probabilities (55% national success
and 45% national failure), but Colaco finds out whether Chocola is a
national success or a national failure before making the decision to market
Chocola nationally or not.
❑ EVPI is a useful upper bound on the value of sample or test market
information.

Your Logo or Name Here

28
Expected Value of Perfect Information (EVPI) (2)

Steps:
1. Calculate expected value with perfect information (EVWPI) → found by
drawing a decision tree in which the decision maker has perfect information
about which state has occurred before making a decision.

2. Calculate expected value of perfect information (EVPI).

EVPI = EVWPI - EVWOI.

Your Logo or Name Here

29
Expected Value of Perfect Information (EVWPI) for
Colaco
1. Expected value with perfect
information (EVWPI) = 315,000

Your Logo or Name Here

30
Expected Value of Perfect Information (EVPI) for Colaco

1. Expected value with perfect

information (EVWPI) = 315,000

2. Expected value of perfect

information (EVPI) =
315,000 – 270,000 = 45,000

Thus, a perfect (one that was

always correct) test marketing
study would be worth $45,000.
that is, no sample or test market
information (no matter how
good) can be worth more than
$45,000.

Your Logo or Name Here

31
Terima Kasih
Yani Herawati
Program Studi Teknik Industri
UNPAR
[email protected]