Value of Information

Honest Decision Tree Scope

Decision trees make assumed probabilities and payoffs visible; they do not learn those inputs.

highlighted = computed this step

Assumed probability

The demand probabilities are assumed. The displayed probability is 1/2 for each demand branch.

p= 1/2p=\ 1/2
Honest scopeExact arithmetic is not the same as learned probabilities.assumed inputs and computed outputspbaseinfoEVPImodel1/2507525

Computed values

The original and information values are $50 and $75.

$50,$75\$50,\$75
Honest scopeExact arithmetic is not the same as learned probabilities.assumed inputs and computed outputspbaseinfoEVPImodel1/2507525

Decision use

The EVPI is $25. Use it as a model consequence, not as evidence that the inputs are correct.

model consequence=$25\text{model consequence}=\$25
Honest scopeExact arithmetic is not the same as learned probabilities.assumed inputs and computed outputspbaseinfoEVPImodel1/2507525

Honesty note

The probabilities and dollar payoffs are assumed inputs. This book folds back a known decision tree; it does not learn probabilities or prove the payoffs will occur. Pixel positions are rounded for layout; every displayed payoff, probability, and expected value is exact.

known inputs, exact rollback\text{known inputs, exact rollback}
Honest scopeExact arithmetic is not the same as learned probabilities.assumed inputs and computed outputspbaseinfoEVPImodel1/2507525