All Tools
F
DataFreeOpen Source
FEATURETOOLS
Automated feature engineering for ML
BSD-3-Clause
ABOUT
Feature engineering is the most time-consuming and domain-expertise-intensive step in ML pipelines, requiring practitioners to manually write dozens of aggregation, transformation, and encoding operations across relational tables. This manual process is slow, inconsistent, and often misses high-value features. Featuretools automates this with deep feature synthesis — automatically creating features by traversing relationships between tables, applying stacking transformations (group-by means, trends, time-since-last), and generating hundreds of candidate features from multi-table datasets in minutes.
INSTALL
pip install featuretoolsINTEGRATION GUIDE
1. Automatically generate hundreds of features from multi-table relational datasets (transactions, customers, products) for fraud detection, credit scoring, and customer churn prediction
2. Combine automated feature engineering with feature selection to build robust ML pipelines that capture complex temporal patterns without manual feature coding
3. Integrate into existing ML workflows as a scikit-learn-compatible transformer for production feature engineering with reproducible, version-controlled feature definitions
TAGS
feature-engineeringautomated-mldata-preparationpythonrelational-datatime-seriesfeature-synthesis