hosmer_lemeshow_test

Note

This page is a reference documentation. It only explains the function signature, and not how to use it. Please refer to the Habitat Guide and Python API guide (v2.0) for usage.

hosmer_lemeshow_test(y_true: ndarray, y_prob: ndarray, n_groups: int = 10) → Tuple[float, float][source]

Perform the Hosmer-Lemeshow calibration test for binary outcomes.

Subjects are grouped into n_groups quantile-based risk groups (right-closed intervals, lowest edge included, mirroring pd.qcut), then the chi-square statistic comparing observed and expected event counts is evaluated against n_groups - 2 degrees of freedom.

Parameters:

  • y_true – Binary outcomes (0/1).

  • y_prob – Predicted event probabilities, aligned to y_true.

  • n_groups – Number of risk groups (classically 10, the decile test).

Returns:

Tuple (statistic, p_value).

Raises:

ValueError – If inputs are misaligned, non-binary, probabilities fall outside [0, 1], or the quantile edges are not unique (too many tied probabilities to form n_groups groups).