pystra.active_learning.estimation.EnrichmentResult

Use it: User guide · Worked example · Theory

class EnrichmentResult(points, mean, std, estimate, probability_band)

Bases: object

Estimator-generated points and probability diagnostics for one fit.

points are row-wise independent-normal coordinates, with mean and std predictions at those points. Their sampling distribution may be conditional or weighted: they are used for selection only. estimate and the ascending probability_band must be computed by the estimator under its sampling measure, never by averaging these pooled predictions. The band describes mean +/- 2 std, not true-model confidence limits. Arrays are copied into read-only buffers; they do not alias sampler state.

Methods