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

Decision Trees 1

Uploaded by

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

Decision Trees 1

Uploaded by

Zwe Zwe
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 15

The illegal parking example using decision trees

Suppose you decide never to buy a parking ticket


If you are caught, then you have to pay a fine of £20
If you are not caught, then you pay nothing
The chance of being caught is 10% i.e. 1 day in 10
What is the Expected Monetary Value (EMV)?
Probability of not being caught = 0.9, outcome £0
Probability of being caught = 0.1, outcome = -£20
EMV = add all probability x outcome = 0.9 x 0 + 0.1 x (-20) = -£2

1
Decision trees

We can use a decision tree to formalise this decision making


process.
Decision trees are useful when there are a series of decisions
to be made.
Two types of node:
Decision nodes
Chance nodes
The other elements are the arcs connecting the nodes.

2
Decision nodes

Represented by squares

The idea is that you choose the best path leaving the decision
node to the right, i.e. the one giving the highest EMV
– £0.60

Buy ticket

NOTE THERE ARE NO PROBABILITIES


ON THE ARCS AFTER A DECISION
NODE

Don’t buy ticket


– £2.00

3
Chance nodes

Represented by circles
The idea is that there is a probability for each arc to the right of
a chance node.
The value at the chance node is the EMV from these arcs.
– £20.00
0.1
Caught NOTE THE PROBABILITIES ON THE
ARCS MUST SUM TO ONE

0.9 NOTE THE RIGHT HAND VALUES ARE


Don’t buy THE OUTCOMES IF THIS HAPPENS

Not caught £0.00

4
Chance nodes

Represented by circles
The idea is that there is a probability for each arc to the right of
a chance node.
The value at the chance node is the EMV from these arcs.
– £20.00
0.1
Caught NOTE THE PROBABILITIES ON THE
ARCS MUST SUM TO ONE
EMV = – £2.00

0.9 NOTE THE RIGHT HAND VALUES ARE


Don’t buy THE OUTCOMES IF THIS HAPPENS

Not caught £0.00

5
Full Decision Tree
Method: Step 1 - Draw options from left to right

Caught

Chance Node
the different paths from it represent
different events happening
Park illegally

Not Caught
?

Decision Node Park legally

6
Full Decision Tree
Method: Step 2 - List Chances & Outcomes, then
calculate the sum of the EMVs

Probability = 0.1 -£20


Caught

Park illegally

Not Caught
? -£0
Probability = 0.9

-£0.60

7
Full Decision Tree
Method: Step 2 - List Chances & Outcomes, then
calculate the sum of the EMVs

Probability = 0.1 -£20


Caught

EMV = -£2
Park illegally

Not Caught
? -£0
Probability = 0.9

-£0.60

Since the Objective is to minimise parking costs in the long run, the answer is:

8
Decision Tree Method
Method: Step 2 - List Chances & Outcomes, then
calculate the sum of the EMVs

Probability = 0.1 -£20


Caught

EMV = -£2
Park illegally

Not Caught
? -£0
Probability = 0.9

Park legally ü -£0.60

Since the Objective is to minimise parking costs in the long run, the answer is:
Park legally to minimise EMV to £0.60/day

9
Example 2 – Inventor

• An inventor has a new idea. To get it to the prototype stage when he could
sell it to a manufacturer he needs to spend £10,000.

• He has no guarantees that he will find a manufacturer, but if he does he


knows from previous experience that he should be able to sell the idea for a
one-off fee of £100,000

• Should he go ahead or not?


– If certain to find a manufacturer.. then yes!
– If completely uncertain then ???

• But if a consultant tells him there’s a 40% chance that he will find a
manufacturer, then using EMV........

10
Example 2 - Inventor
Method: Step 1 - Draw options from left to right

THINK ABOUT WHAT HAPPENS FIRST


THEN WHAT HAPPENS NEXT, etc.

Don’t Develop

Sells Idea 40%


?

Develop

This option incurs


development costs of -
£10k for all outcomes

Doesn’t Sell Idea 60%

11
Example 2 - Inventor
Method: Step 2 - List Chances & Outcomes, then calculate the sum of the EMVs

Don’t Develop

£0

0.4 Sells Idea


? £90k Why isn’t this £100k?
gets £100k

This option incurs


development costs of -
£10k for all outcomes

Doesn’t Sell Idea


-£10k
0.6

12
Example 2 - Inventor
Method: Step 2 - List Chances & Outcomes, then calculate the sum of the EMVs

Don’t Develop

£0

0.4 Sells Idea


? £90k

£30k From 0.4 × 90k + 0.6 × -10k

Doesn’t Sell Idea


-£10k
0.6

13
Example 2 - Inventor
Method: Step 2 - List Chances & Outcomes, then calculate the sum of the EMVs

Don’t Develop

£0

0.4 Sells Idea


? £90k
£30k
Develop ü

£30k

Answer: Develop the


idea to maximise long
term average profits
Doesn’t Sell Idea
with an EMV of £30k
-£10k
0.6

14
When is EMV sensible?
• When the individual financial outcomes are not too large
• If there is a number of similar events e.g. you will be
working many nights in the restaurant or bar, you will be
parking on many occasions
So often appropriate for business decisions

For individuals, outcomes may be too large


(and behavioural economics has shown that people are
often irrational when making financial decisions)
15

You might also like