_scores

cultivars.engine._core._scores = <module 'cultivars.engine._core._scores' from '/home/runner/work/cultivars/cultivars/src/cultivars/engine/_core/_scores.py'>

exact functionals of a predictive sample.

Each function here evaluates a strictly proper scoring rule – or the probability integral transform that calibration reads – directly from a predictive sample, using exact finite-sample identities rather than density estimates wherever the mathematics allows. The orientation convention is stated once: every score is negatively oriented, smaller is better, so rules and losses can share tables without a legend. Draws are axis zero everywhere.

Type:

Scoring primitives