An open source framework for automated feature engineering

Developers at MIT and Spanish bank BBVA used Featuretools to build features to train better fraud detection models.  Read more >

What is Featuretools?

Deep Feature Synthesis

Implements DFS for automated feature engineering.

It works to prepare raw relational and transactions datasets for machine learning or predictive modeling.

Feature Primitives

Includes a collection of reusable feature engineering functions for a wide range of domains.

Allows for user-defined primitives to encourage automation, reuse across projects, and community collaboration.


Easily integrated in to existing production data pipelines.

Designed to work with common frameworks like Pandas for data preparation or scikit-learn for machine learning.

Used by

Stay up to date

Get the latest Featuretools news