0% found this document useful (0 votes)
71 views

Influence Diagrams and Decision Trees

The document describes three examples of using influence diagrams and decision trees to model decision making under uncertainty. 1) The first example models a sports betting decision between betting on Team A, Team B, or not betting at all. The decision tree and influence diagram both show that betting on Team A has the highest expected value of $2. 2) The second example models investing $2,000 in a risky business or saving it. The decision tree shows the business has a 50% chance of a $5,000 profit but a 50% chance of losing all $2,000, while saving yields a guaranteed 10% return of $2,200. 3) The third example models an oil wild

Uploaded by

Luis E G
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
71 views

Influence Diagrams and Decision Trees

The document describes three examples of using influence diagrams and decision trees to model decision making under uncertainty. 1) The first example models a sports betting decision between betting on Team A, Team B, or not betting at all. The decision tree and influence diagram both show that betting on Team A has the highest expected value of $2. 2) The second example models investing $2,000 in a risky business or saving it. The decision tree shows the business has a 50% chance of a $5,000 profit but a 50% chance of losing all $2,000, while saving yields a guaranteed 10% return of $2,200. 3) The third example models an oil wild

Uploaded by

Luis E G
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 26

Influence Diagrams and Decision Trees

Example 1

Your decision is to bet on Team A, Team B or not at all. The bet costs $1.
The probability of Team A winning is 37.5% and Team B winning is 62.5%.
If Team A wins you get $8 and if Team B wins you get $3. Decide on what
team to bet on.

Since this is a fairly simple problem, we use the table we developed earlier
and create the payoff table.

Expected Value Team A wins Team B wins  Exp Value MaxEV Decision


Team A 7 ‐1 2 Team A
Team B ‐1 2 0.875
Nothing 0 0 0 2
Probability 0.375 0.625      

The payoff table includes the cost of the bet, which is -$1. We try to do this
problem using decision tree and influence diagram.

Decision Tree

Step 1: Click on the decision tree icon and identify the cell you want the tree
to start. Here cell A20.

Page 1
Name the tree Sports betting and click OK. Click on the end node, which is a
triangle and add a decision node with 3 decision alternatives.

Name the alternatives as Bet Team A, Bet Team B and No Betting. The
value or cost of betting either team A or team B is -$1 and no cost if you do
not bet.

Page 2
Add a chance node at the end of Bet Team A decision alternative and add
two branches to the chance node with correct probabilities and payoffs.

When you bet on Team A, if team A wins, you receive $8 and if team B
wins you get nothing. The probabilities of team A and B winning are 37.5%
and 62.5% respectively.

Page 3
Instead of creating a sub tree for the other alternative which is Bet Team B,
we can copy the sub tree and paste it at the appropriate place. Just right click
on the chance node and click on copy sub tree and paste the sub tree at the
end of the decision alternative Bet Team B.

Page 4
The payoff values are not correct. You can directly click on the cells and
change those values. Instead of 8 and 0, it should be 0 and 3. This means, if
you bet on team B, if team A wins, you get nothing and if team B wins, you
receive $3. The resulting tree will look like this.

Page 5
The maximum expected value is $2 and the decision is to bet on Team A.
The correct decision path will be indicated by TRUE and others with the
word FALSE. There is no reason to continue the No Betting decision
alternative.

Influence Diagram

Click on the Influence diagram/node icon.

After you assigned a cell and click OK, Change the name to Sports Betting
Influence Diagram and hit ok.

Add another node and make this as a decision node.

Page 6
Click on the outcomes and add one more outcome and type, Bet Team A,
Bet Team B and No Betting.

Change the name of the decision block to Sports Bet and click on Exit to
Value Table. Betting on Team A and Team B costs $1 and No Betting does
not cost any money.

Page 7
Create a chance node with two outcomes, Team A wins and Team B wins,
with probabilities and payoff values. Change the name of the chance node to
Betting Results, with two outcomes, Team A wins and Team B wins. Click
Exit to Value Table.

Page 8
Connect the decision node to the payoff node and the chance node to the
Payoff node. The resulting diagram will look like this.

Click on the Payoff node and click the Node settings. Click Exit to Value
Table.

Page 9
You need to insert the values of payoff as formulas. For Bet Team A and
Team A wins the payoff is the addition of those two cells. For Bet Team A
and if Team B wins, the value is only the Bet Team A. For Bet Team B and
if Team B wins, the value is the addition of the two cells.

Expected 
Value  2
Std. Deviation  3.872983346
Minimum  ‐1
Maximum  7

The expected value of this analysis is $2 and the payoff minimum is -$1(lose
the dollar bet) and maximum is $7 ($8 win - $1 bet).

By right clicking on the Decision node and clicking on Model and clicking
on Convert to Decision Tree, this software automatically creates the decision
tree.

Page 10
37.5% 37.5%
Team A wins
8 7
TRUE Betting Results
Bet Team  A
‐1 2
62.5% 62.5%
Team B wins
3 ‐1
Sports Bet
Converted Sports Betting Influence Diagram
2
37.5% 0.0%
Team A wins
8 ‐1
FALSE Betting Results
Bet Team  B
‐1 0.875
62.5% 0.0%
Team B wins
3 2
37.5% 0.0%
Team A wins
8 0
FALSE Betting Results
No Betting
0 0
62.5% 0.0%
Team B wins
3 0

Example 2

Choice is whether to invest in business or savings. If business is chosen,


then the outcome of the chance is a wild success or a flop. For a $2,000
investment, you can end up with a 50-50 chance of a wild success of $5,000,
or in flop, you lose everything. In savings you get a 10% interest and your
balance is $2,200 for sure. What is your decision?

Page 11
The decision tree has two branches – business investment or savings. In the
business investment there are two chance states of nature, wild success or
flop. For the savings, the outcome is 10% more than the investment.

50.0% 50.0%
Wild Success
5000 3000
TRUE Chance
Business
‐2000 500
50.0% 50.0%
Flop
0 ‐2000
Decision
Investmetn
500
50.0% 0.0%
Branch #1
2200 200
FALSE Chance
Savings
‐2000 200
50.0% 0.0%
Branch #2
2200 200

Page 12
Page 13
Example 3

An oil wildcatter must decide whether to drill or don’t drill. He is uncertain


whether the hole is dry, wet or soaking. The cost of drilling is $70,000.The
subjective probabilities of the various states are 0.5 for dry, 0.3 for wet and
0.2 for soaking. The payoffs for these three situations are $0, $120,000 and
$270,000. We assume that the wildcatter’s utility function is a linear
function of the monetary profits, i.e., the wildcatter is risk neutral (we will
discuss this later in Lesson 6).

At a cost of $10,000 the wildcatter could take seismic soundings that will
help determine the geological structure of the site. The soundings will
disclose whether the terrain below has no structure (which is bad), open

Page 14
structure (which is so-so) or closed structure (which is very hopeful). The
experts have provided that there is a 41% chance there will be no structure,
35% chance there is a open structure and a 24% chance there is a closed
structure. They can also estimate the probabilities of finding oil under these
three structural conditions.

Dry Wet  Soaking 


No Structure 73% 22% 5%
Open Structure 43% 34% 23%
Closed Structure 21% 37% 42%

Create a decision tree and analyze the situation. What decision should the oil
wildcatter make?

Page 15
Page 16
Page 17
Page 18
Page 19
Page 20
Page 21
Page 22
Page 23
Page 24
Page 25
Page 26

You might also like