ROC#

class interpret.perf.ROC(model, feature_names=None, feature_types=None)#

Produces ROC curves.

Initializes class.

Parameters:

  • model – model or prediction function of model (predict_proba for classification or predict for regression)

  • feature_names – List of feature names.

  • feature_types – List of feature types.

explain_perf(X, y, name=None)#

Produce ROC curves.

Parameters:

  • X – Numpy array for X to compare predict function against.

  • y – Numpy vector for y to compare predict function against.

  • name – User-defined explanation name.

Returns:

An explanation object.