Basic Decision Tree
Basic Decision Tree
Manahan Siallagan
Problems of Choice
C1
O1
C2
O2
e
b
e
Th
2
ay
w
st
D : a decision maker
C : possible courses of action
(alternatives)
O1 : desirable outcome;
O2 : undesirable outcome
consequences (Strategic/Interactive
Decision).
I cant see
customers ? Competitors?
Investing in securities :
How is the economy?
How about interest rates?
location?
How much?
E1,1
p1,1
O1
E1,2
E2,1
p1,2 O
2
p2,1 O1
E2,2
p2,2
O2
D : a decision maker
C : possible courses of action (alternatives)
O1 and O2 : possible outcomes/consequences/payoffs
Ei,j : Events (State of Natures/SON)
6
pi,j : probabilities
choose;
Situation controllable by decision maker.
Alternatives of actions
7
by decision maker;
Uncontrollable by decision maker;
Decision maker only has information about
probability of each event no complete
information.
E1
E2
p1
p2
An action
E3
8
p3
Events
Building Decision
Tree
1. Identify what decisions should be made
by DM;
first
decision, and
decisions to be made?
next
decision;
3. Draw decision node and event node
(SON);
4. Complete information about probabilities;
5. Complete information about payof.
9
Goferbroke
Company(1)
Max Flyer is the founder of and sole owner of the
Goferbroke Company(2)
On the other hand, if the tract does contain oil, the
State of Nature
Oil
Dry
$700
-$100
$90
$90
Max in Row
700
90
Maximax
(total pessimist)
Identify the minimum payoff from any SoN for each
decision alternatives
Find the maximum of these minimum payoffs and
choose the corresponding decision alternative
Weakness : abandoning prior probability and
abandoning other payoff beside only the maximin.
Alternative
Drill for oil
Sell the land
State of Nature
Oil
Dry
$700
-$100
$90
$90
Min in Row
-$100
90
Maximin
prior probability;
Choose the decision alternative that has
the largest payoff for this state of nature.
Weakness: abandoning payoff, that
actually payoff maybe very big.
Alternative
Drill for oil
Sell the land
Prior Probability
Step 1 : Maximum
Step 2 : Maximum
Advantage :
Considering entire information (alternatives, payoffs, and prior
probabilities);
In long term, if the decision occur sequential, then this criteria
will resulting payoff that mostly probably happen.
Alternative
Drill for oil
Sell the land
Prior Probability
100
90
Maximum
DT of Goferbrokes Case
Decision:
Drill or Sell the Land
SON
Oil or Dry
Decision nodes
Drill
-100
Sell
90
DT of Goferbrokes Case
Event nodes
0.25
Oil
800
Drill
-100
Sell
90
Dry
0,75
0
DT of Goferbrokes Case
Payof
0.25
Oil
Drill
-100
Sell
90
800
Dry
0,75
0
700
-100
90
DT of Goferbrokes Case
Payof
0.25
Oil
800
Drill
-100
100
100
0,75
0
Sell
90
Dry
90
700
-100
90
Expected payoff
= MAX [100,90]
= 100
Action: Drill
100=(0.25*700) + (0.75*(-100))
Decision Analysis:
New Information or Posterior
Probability
23
Posterior
probability
24
Assessing Probability
There are three approaches to assessing the
X
number of ways the event can occur
T
total number of elementary outcomes
3. subjective probability
an individual judgment or opinion about the probability of
occurrence
Type
Color
Red
Black
Total
Non-Ace
24
24
48
Total
26
26
52
Ace
Type
Color
2
2
4
52 52 52
Red
Black
Total
Non-Ace
24
24
48
Total
26
26
52
Ace
P(A and B)
P(A | B)
P(B)
P(A and B)
P(B | A)
P(A)
The conditional
probability of A
given that B has
occurred
The conditional
probability of B
given that A has
occurred
Conditional Probability
Example
Conditional Probability
Example
(continued
Of the cars on a used car lot, 70% have air conditioning
) (AC) and
40% have a CD player (CD).
20% of the cars have both.
CD
No CD
Total
AC
0.2
0.5
0.7
No AC
0.2
0.1
0.3
Total
0.4
0.6
1.0
0.2857
P(AC)
0.7
Conditional Probability
Example
(continued
)
Given AC, we only consider the top row (70% of the cars). Of
these, 20% have a CD player. 20% of 70% is about 28.57%.
CD
No CD
Total
AC
0.2
0.5
0.7
No AC
0.2
0.1
0.3
Total
0.4
0.6
1.0
0.2857
P(AC)
0.7
Given AC or no
AC:
Ha
All
Car
s
)=
C
A
P(
C 0.7
A
s
Do
e
not s
hav
AC e
P(A
0.3 C)=
Do e
s
not
hav
e
CD
.5
.7
.2
.3
D
C
Has
Do e
s
not
hav
e
CD
.1
.3
Given CD or no
CD:
)=
D
P(C
D 0.4
C
s
Ha
All
Car
s
Do
e
not s
hav
CD e
P(C
0.6 D)=
Do e
s
not
hav
e
AC
.2
.4
.5
.6
C
A
Has
Do e
s
not
hav
e
AC
.1
.6
(continued
)
P(CD and AC) =
0.2
P(CD and AC) =
0.2
Bayes Theorem
Bayes theorem is used to revise previously calculated
probabilities after new information is obtained
P(A | B i )P(B i )
P(B i | A)
P(A | B 1 )P(B 1 ) P(A | B 2 )P(B 2 ) P(A | B k )P(B k )
where:
Goferbrokes Case
Continued
Survey by geologist will provide more
Events:
Do Survey
Drill or Sell
Oil
Dry
30000
Max`s
Experience
P(state) prior; which is P(Oil)=0.25 & P(Dry)=0.75;
P (finding|state) being known based on Maxs
experiences; which is
P(FSS|Oil)=0.6,
P(USS|Oil)=0.4,
P(FSS|Dry)=0.2, and
P(USS|Dry)=0.8
Which:
Contingency
table
FSS
USS
Oil
0.15
0.1
0.25
Dry
0.15
0.6
0.75
0.3
0.7
P(FSS) = P(FSS&Oil)+P(FSS&Dry)
= 0.3
FSS
USS
Oil
0.15
0.1
0.25
Dry
0.15
0.6
0.3
0.7
P(FSS)
43
0.3
P(USS)
670
Drill
800
-100
-15,714
0,7
Unfavorable
0,857
Dry
-30
670
P(Dry|USS)
-130
Expected payoff
= MAX [123,100]
Do S urvey
= 123
P(Oil|USS)
-130
60
Sell
60
90
60
0,5
123
P(FSS)
670
Drill
800
-100
270
0,3
Favorable
0,5
Dry
670
P(Dry|FSS)
-130
1
0
P(Oil|FSS)
Oil
-130
270
Sell
123
60
Drill
-100
100
90
60
0,25
Oil
P(Oil)
800
700
0,75
Dry
P(Dry)
-100
No S urvey
1
0
700
-100
100
Sell
90
90
90
Posterior Probability
Given:
P(state)=prior probability: P(oil) and P(dry)
P(finding|state) = Maxs experience on probabilities of finding (FSS or USS)
could occur if some SoN (oil or dry) has been already happened.
P(FSS|oil)
0.25*0.6=0.15
Oil and FSS
0.15/0.3=0.5
Oil, diket FSS
P(oil|FSS)
P(state|finding)
Prob. posterior
P(USS|oil)
0.1/0.7=0.14
Oil, diket USS
P(oil|USS)
0.75*0.2=0.15
Dry and FSS
P(dry)
0.15/0.3=0.5
Dry, diket FSS
P(dry|FSS)
0.75*0.8=0.6
Dry and USS
0.6/0.7=0.86
Dry, diket USS
P(dry|USS)
Thank You
47
Exercise
Von Holt was a general marketing manager. He was assigned