seismometer.api.plot_model_evaluation

seismometer.api.plot_model_evaluation(cohort_dict, target_column, score_column, thresholds, per_context=False)

Generates a 2x3 plot showing the performance of a model.

This includes the ROC, recall vs predicted condition prevalence, calibration, PPV vs sensitivity, sensitivity/specificity/ppv, and a histogram.

Parameters:

• cohort_dict (dict[str, tuple[Any]]) – dictionary of cohort columns and values used to subselect a population for evaluation

• target_column (str) – target column

• score_column (str) – score column

• thresholds (list[float]) – thresholds to highlight

• per_context (bool, optional) – if scores should be grouped, by default False

Returns:

an html visualization of the model evaluation metrics

Return type:

HTML