ImportanceResult

Defined in fynance.research

class ImportanceResult(importances, stds, baseline, n_folds, feature_names=None)

Bases: object

Result of walk_forward_mda.

Parameters:

importancesnumpy.ndarray, shape (n_features,)

Mean out-of-sample score drop per permuted feature, averaged across folds and repeats. Higher means more important (permuting the feature hurts the score more).

stdsnumpy.ndarray, shape (n_features,)

Standard deviation of the score drop across folds x repeats.

baselinefloat

Mean (unpermuted) out-of-sample score across folds.

n_foldsint

Number of walk-forward folds evaluated.

feature_nameslist of str, optional

Feature labels aligned with importances/stds, if provided.