API

Base Classes

BaseVariableImportance()	Base class for variable importance methods.
BasePerturbation(estimator, method, loss, ...)	Abstract base class for model-agnostic variable importance measures using perturbation techniques.
BasePerturbationCV(estimators, cv[, ...])	Base class for perturbation methods with cross-validation.
GroupVariableImportanceMixin([features_groups])	Mixin class for adding group functionality to variable importance methods.

Feature Importance Classes

LOCO(estimator, method, loss[, ...])	Leave-One-Covariate-Out (LOCO) algorithm
LOCOCV(estimators, cv[, statistical_test, ...])	Leave-One-Covariate-Out (LOCO) algorithm with Cross-Validation.
CFI(estimator, method, loss, n_permutations)	Conditional Feature Importance (CFI) algorithm.
CFICV(estimators, cv[, statistical_test, ...])	Conditional Feature Importance (CFI) algorithm with Cross-Validation.
CluDL(clustering[, desparsified_lasso, ...])	Clustered inference with desparsified lasso.
EnCluDL(desparsified_lasso, clustering[, ...])	Ensemble clustered inference with desparsified lasso.
PFI(estimator, method, loss, n_permutations)	Permutation Feature Importance algorithm
PFICV(estimators, cv[, statistical_test, ...])	Permutation Feature Importance (PFI) algorithm with Cross-Validation.
D0CRT(estimator[, method, estimated_coef, ...])	Implements distilled conditional randomization test (dCRT) without interactions.
ModelXKnockoff([estimator, max_iter, ...])	Model-X Knockoff
DesparsifiedLasso([estimator, centered, ...])	Desparsified Lasso Estimator (also known as Debiased Lasso)

Feature Importance functions

cfi_importance(estimator, X, y[, method, ...])	Conditional Feature Importance (CFI) algorithm.
d0crt_importance(estimator, X, y[, cv, ...])	Implements distilled conditional randomization test (dCRT) without interactions.
desparsified_lasso_importance(X, y[, ...])	Desparsified Lasso Estimator (also known as Debiased Lasso)
model_x_knockoff_importance(X, y[, ...])	Model-X Knockoff
loco_importance(estimator, X, y[, method, ...])	Leave-One-Covariate-Out (LOCO) algorithm
pfi_importance(estimator, X, y[, method, ...])	Permutation Feature Importance algorithm

Visualization

PDP(estimator[, feature_names])

Partial Dependence Plot (PDP) visualization.

Samplers

ConditionalSampler([model_regression, ...])
GaussianKnockoffs([cov_estimator, tol])	Generator for second-order Gaussian variables using the equi-correlated method.

Helper Functions

quantile_aggregation(pvals[, gamma, adaptive])	Implements the quantile aggregation method for p-values.
reid(beta_hat, residual[, tolerance, ...])	Residual sum of squares based estimators for noise standard deviation estimation.
nadeau_bengio_ttest(a, popmean, test_frac[, ...])	One-sample t-test with Nadeau & Bengio variance correction.