Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

25 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

xuplift is a library for explainable uplift modeling. It uses linearized kernel feature maps to estimate treatment effects with both speed and mathematical rigor. Instead of computing a massive $N \times N$ kernel matrix, xuplift selects landmark points to project data into a finite-dimensional feature space.

Supported Models

  • Regressor: Kernel-based Ridge regressor for outcome and residual modeling.
  • Classifier: Kernel-based Logistic classifier for precise propensity score estimation.

Supported Meta-Learners

  • DRClassifier, DRRegressor: Doubly robust estimator combining propensity scores and outcome models.
  • GRClassifier, GRRegressor: Generalized R-learner supporting both continuous and binary treatments.
  • MRegressor: Modified covariates learner optimized for randomized controlled trials (RCT).
  • PWRegressor: Propensity score weighted learner using inverse probability weighting.
  • RClassifier, RRegressor: Residual learner minimizing an R-objective via residual-on-residual regression.
  • SClassifier, SRegressor: Single learner treating treatment assignment as a standard feature.
  • TClassifier, TRegressor: Two learner approach fitting independent models for each group.
  • XClassifier, XRegressor: Cross learner optimized for significantly unbalanced treatment groups.

Installation

pip install xuplift

About

Explainable uplift modeling via linearized kernel feature maps

Topics

Resources

Stars

9 stars

Watchers

0 watching

Forks

Releases

Contributors

Languages