In: Statistics and Probability
In your own words, explain how to obtain the “expected value of perfect information” for any payoff table, which has probabilities associated with each state of nature. Then, provide an example
Expected Value of Prefect Information (EVPI) is calculated as the difference between Expected Value Under Uncertainty(EVUU). and Expected Value using a Perfect Prediction (EVPP). It is the value of having perfect information without any uncertainty . For any payoff table which has probabilities associated with each state of nature the following steps are used for the purpose of calculating the Expected Value of Perfect Information (evpi).
1. First the Expected Monetary value (EMV) is calculated for each alternative decision or action.The expected monetary value for an alternative course of action is calculated by adding the product of payoffs associated with each state of nature and their respective probabilities.In case of a decision criterion like maximin the EMV for each alternative course of action will be the minimum value for that course of action.
2. Once the EMV is calculated for each alternative course of action EVUU needs to be ascertained from the EMV's calculated. In case the probabilities for the states of nature are given with no specific decision criterion the EVUU is taken to be the maximum of the EMV's calculated. In case of decision criterion like maximin the maximum of the minimum value (EMV's) is taken to be the EVUU.
3. Next the EVPP needs to be calculated. For the purpose of calculating EVPP the maximum value for each state of nature is identified. Then the values for each state of nature are multiplied with their respective probabilities. In case the probabilities for the states of nature are not given each state of nature is considered to be equally likely. The product of the maximum payoffs and probabilities are then added up to obtain EVPP.
4. Finally the EVPI is calculated by subtracting EVUU from EVPP. Then EVPI=EVPP-EVUU