MorrisSensitivity#

Link to Algorithm description: Morris Sensitivity Analysis

class interpret.blackbox.MorrisSensitivity(model, data, feature_names=None, feature_types=None, sampler=None, **kwargs)#

Method of Morris for analyzing blackbox systems. If using this please cite the package owners as can be found here: SALib/SALib

Morris, Max D. “Factorial sampling plans for preliminary computational experiments.” Technometrics 33.2 (1991): 161-174.

Initializes class.

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

  • data – Data used to initialize LIME with.

  • feature_names – List of feature names.

  • feature_types – List of feature types.

  • sampler – A SamplerMixin derrived class that can generate samples from data

  • **kwargs – Kwargs that will be sent to SALib.analyze.morris.analyze

explain_global(name=None)#

Provides approximate global explanation for blackbox model.

Parameters:

name – User-defined explanation name.

Returns:

An explanation object, visualizes dependence plots.