Counterfactual explanations¶
-
class
optbinning.scorecard.Counterfactual(scorecard, special_missing=False, n_jobs=1, verbose=False)¶ Bases:
optbinning.scorecard.counterfactual.base.BaseCounterfactualOptimal counterfactual explanations given a scorecard model.
- Parameters
scorecard (object) – A
Scorecardinstance.special_missing (bool (default=False)) – Whether the special and missing bin are considered as valid counterfactual values.
n_jobs (int, optional (default=1)) – Number of cores to run the optimization solver.
verbose (bool (default=False)) – Enable verbose output.
-
display(show_only_changes=False, show_outcome=False)¶ Display the generatedcounterfactual explanations.
- Parameters
show_only_changes (boolean (default=False)) – Whether to show only changes on feature values.
show_outcome (boolean (default=False)) – Whether to add a column with the scorecard outcome. If
outcome_typeis “binary” or “probability”, the estimated probability of the counterfactual is added.
- Returns
counterfactuals – Counterfactual explanations.
- Return type
pandas.DataFrame
-
fit(X)¶ Fit counterfactual. Compute problem data to generate counterfactual explanations.
- Parameters
X (pandas.DataFrame (n_samples, n_features)) – Training vector, where n_samples is the number of samples.
- Returns
self – Fitted counterfactual.
- Return type
-
generate(query, y, outcome_type, n_cf, method='weighted', objectives=None, max_changes=None, actionable_features=None, hard_constraints=None, soft_constraints=None, priority_tol=0.1, time_limit=10)¶ Generate counterfactual explanations given objectives and constraints.
- Parameters
query (dict or pandas.DataFrame) – Input data points for which a single or multiple counterfactual explanations are to be generated.
y (int or float) – Desired outcome.
outcome_type (str) – Desired outcome type. Supported outcome types are “binary”, “probability” and “continuous”.
n_cf (int) – Number of counterfactuals to be generated.
method (str (default="weighted")) – Multi-objective optimization method. Supported methods are “weighted” and “hierarchical”.
objectives (dict or None (default=None)) – Objectives with their corresponding weights or priorities, depending on the method.
max_changes (int or None (default=None)) – Maximum number of features to be changed. If None, the maximum number of changes is half of the number of features.
actionable_features (array-like or None (default=None)) – List of actionable features. If None. all features are suitable to be changed.
hard_constraints (array-like or None (default=None)) – Constraint to be enforced when solving the underlying optimization problem.
soft_constraints (dict or None (default=None)) – Constraints to be moved to the objective function as a penalization term.
priority_tol (float, optional (default=0.1)) – Relative tolerance when solving the multi-objective optimization problem with
method="hierarchical".time_limit (int (default=10)) – The maximum time in seconds to run the optimization solver.
- Returns
self – Generated counterfactuals.
- Return type
-
get_params(deep=True)¶ Get parameters for this estimator.
- Parameters
deep (bool, default=True) – If True, will return the parameters for this estimator and contained subobjects that are estimators.
- Returns
params – Parameter names mapped to their values.
- Return type
dict
-
information(print_level=1)¶ Print overview information about the options settings and statistics.
- Parameters
print_level (int (default=1)) – Level of details.
-
set_params(**params)¶ Set the parameters of this estimator.
The method works on simple estimators as well as on nested objects (such as
Pipeline). The latter have parameters of the form<component>__<parameter>so that it’s possible to update each component of a nested object.- Parameters
**params (dict) – Estimator parameters.
- Returns
self – Estimator instance.
- Return type
estimator instance
-
property
status¶ The status of the underlying optimization solver.
- Returns
status
- Return type
str