koaning/scikit-lego

Provides domain-specific transformers, classifiers, and metrics designed for fairness constraints, time-series handling, and outlier detection—implementing scikit-learn's estimator API standards to ensure seamless pipeline integration. Includes specialized components like demographic parity classifiers, quantile regression, Gaussian mixture models, and grouped predictors for stratified data, alongside utility functions for feature engineering and model selection with temporal gaps.

1,383 stars and 55,486 monthly downloads. Used by 1 other package. Available on PyPI.