libuplift.meta.dr_learner

The DR-learner model from Kennedy 2004.

Classes

DRLearnerBase	Base class for uplift meta estimators.
DRLearnerUpliftRegressor	Mixin class for all uplift regression estimators in
DRLearnerUpliftClassifier	The classifier works by treating the class variable as 0/1 real

Module Contents

class libuplift.meta.dr_learner.DRLearnerBase(base_estimator=LinearRegression(), mean_estimator=None, cv=2)

Bases: libuplift.meta.base.UpliftMetaModelBase

Base class for uplift meta estimators.

Checks input consistency, builds classifiers on subsets of data.

Derived classess need to overwride the _get_model_names_list and _iter_training_subsets methods. The predict method needs to be implemented as well.

predict(X)

fit(X, y, trt, n_trt=None, sample_weight=None, *, y_stratify=None)

class libuplift.meta.dr_learner.DRLearnerUpliftRegressor(base_estimator=LinearRegression(), mean_estimator=None, cv=2)

Bases: libuplift.base.UpliftRegressorMixin, DRLearnerBase

Mixin class for all uplift regression estimators in libuplift.

class libuplift.meta.dr_learner.DRLearnerUpliftClassifier(base_estimator=LinearRegression(), mean_estimator=None, cv=2)

Bases: libuplift.base.UpliftClassifierMixin, DRLearnerBase

The classifier works by treating the class variable as 0/1 real target and using a DRLearnerUpliftRegressor.

The main difference is that stratification takes the target variable into account.

fit(X, y, trt, n_trt=None, sample_weight=None, *, y_stratify=None)

predict(X)