Skip to content

Python Analysis API

The Python facade exposes DecisionAnalysis while the stable numerical boundary uses selected Rust kernels.

Construct it with a ValueArray of net benefits and a ParameterSet of PSA samples. The primary methods are:

  • evpi() — expected value of perfect information.
  • evppi(parameter_names=None) — expected value of partial perfect information for selected parameters.

EVSI is available through the functions in voiage.methods.sample_information; it is not a DecisionAnalysis method.

The opt-in voiage.contracts.run_evpi() route returns a typed result envelope with decision identifiers, numerical policy, diagnostics, and provenance.