liveBook · Manning

Terminal analysis

As explained in Timeless Algorithms, The Foundational Ideas

Terminal analysis is decision analysis conducted after information is observed, using the resulting beliefs to choose an act.

Terminal analysis answers what to do once evidence is known: use posterior probabilities to compare the expected utility of each available act. It differs from preposterior analysis, which evaluates whether to obtain information before its result is known. section 7.2.3 section 7.2.2

a*= arg⁡maxa[∑θ u(a, θ)p(θ | data)]

For each act a, weight its utility in each state θ by that state’s posterior probability given the data, then choose the act with the greatest total. section 7.2.2

Treating after a test

After observing a test result, a physician compares treating and not treating using the resulting probabilities of disease and no disease, then chooses the act with higher expected utility. section 7.2.3 section 7.2.2

How Timeless Algorithms, The Foundational Ideas covers it

How deep it goes
Calculates expected utility from posterior state probabilities, while leaving the value of acquiring future information to preposterior analysis.
Examples
Uses a physician’s choice to treat or not treat after observing a test result.
What it assumes
Requires specified acts, states, current beliefs, and utilities for possible act–state outcomes.
What you can do after
Apply posterior probabilities and utilities to select a preferred act after evidence is observed.
How it connects
Connects expected utility and Bayesian inference with preposterior analysis and value-of-information analysis.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Timeless Algorithms, The Foundational Ideas. The section links point to the parts of the book it draws on.