Expected values for Perfect Information

1
Expected values for Perfect Information
2
Website for information is as follows

• http//en.wikipedia.org/wiki/Expected_value_of_per
  fect_information
• We will now use our classroom example to compute
  based on the above note next.

3
Formulae

• EVPI A B
• A expected value with perfect information
• B expected value without perfect information

4
example

• Consider our lecture example

5
Calculation of A value

• A the best of each outcome x their prob.
• The best of outcomes

Best outcome 100,000
30,000 A 0.6 x 100,000 0.4 x 30,000 72,000
6
Calculation of B value

• B we select the max value of each given below
• Outcome of each event

0.6(50000) 0.4 (30,000) 42,000 0.6(100,000
-0.4(40,000) 44,000 0.6(30,000) 0.4(10,000)
20,000
The max value for all computed value 44,000
7
EVPI

• EVPI A B
• 72,000 44,000
• 28,000

8
More examples!

• You can refer to the website provided for more
  example.
• We show these steps next

9
Consider
0.5 for increase 0.3 for equal 20 decrease
Option 1 1,500 300 -800
Option 2 900 600 -200
Option 3 500 500 500
EVPI A B A expected value with perfect
information B expected value without perfect
information For A The optimal values for each
value are
0.5 for increase 0.3 for equal 20 decrease
Option 1 1,500 300 -800
Option 2 900 600 -200
Option 3 500 500 500
Max value 1,500 600
500 Thus A 0.5 x 1,500 0.3 x 600 0.2 x 500
1030
10
For B we compute the expected values for each
column first, and then select the max as below
0.5 for increase 0.3 for equal 20 decrease Max value
Option 1 1,500 300 -800 0.5x1,5000.3x3000.2x(-800) 680
Option 2 900 600 -200 0.5x9000.3x6000.2x(-200) 590
Option 3 500 500 500 0.5x5000.3x5000.2x500 500
Thus B max of all max value of above
680 EVPI A B 1030 680 350
11

• These steps also applied and worked for all
  examples shown in the tut as well.
• Good luck!