feature_engine and featuretools
These are complementary tools: featuretools automates the creation of new features from relational data, while feature_engine provides a comprehensive toolkit for selecting, engineering, and transforming features—they're typically used together in a feature engineering pipeline.
About feature_engine
feature-engine/feature_engine
Feature engineering and selection open-source Python library compatible with sklearn.
Provides 70+ specialized transformers for imputation, encoding, discretization, outlier handling, variable transformation, and feature selection—each following sklearn's fit/transform API for seamless pipeline integration. Covers domain-specific engineering including datetime extraction, text feature generation, lag/window features for time series, and statistical selection methods like mutual information and PSI-based filtering. Handles end-to-end preprocessing workflows from missing data to feature creation and elimination across tabular, temporal, and text data.
About featuretools
alteryx/featuretools
An open source python library for automated feature engineering
Implements Deep Feature Synthesis (DFS), an algorithm that automatically generates features from relational, multi-table datasets by applying aggregation and transformation primitives across entity relationships and time windows. Supports distributed computation via Dask for parallel feature generation, extensible primitive libraries (including NLP and premium options), and can handle timestamped transactions to create temporal features.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work