featuretools/featuretools
Featuretools
Featuretools is a Python data science library and automated feature engineering framework designed to create predictive features from multiple related datasets. It automates the data preparation and transformation steps required for machine learning models through deep feature synthesis.
The library enables the automatic generation of comprehensive feature tables by applying recursive transformations to relational data. It supports the transformation of unstructured text into structured numeric features and allows users to define custom primitives to extend the synthesis process with specific logic.
The system handles large-scale data pipelines by distributing feature computation across multiple CPU cores or clusters. This scaling is achieved through integrations with distributed computing frameworks to process massive volumes of data.
Features
- Feature Engineering - Provides an automated framework for transforming raw relational data into meaningful predictive features for machine learning.
- Deep Feature Synthesis - Implements deep feature synthesis to automatically generate predictive variables from relational datasets.
- Modular Primitives - Encapsulates feature engineering logic into modular, reusable primitives for consistent data synthesis.
- Recursive Expansion - Builds complex predictive features through a recursive chain of operations on related data.
- Relational Preprocessing - Automatically synthesizes features from multiple related tables to prepare complex relational data for predictive models.
- Tabular Feature Engineering - A Python framework for converting raw tabular and relational data into model-ready formats through automated synthesis.
- Data Science Libraries - A Python library providing essential tools for the automated data preparation and transformation steps in data science.
- Relational Mappings - Uses relational mappings and foreign keys to automate the creation of aggregations and joins across datasets.
- Automated Feature Tables - Synthesizes a single comprehensive table of features by applying automated synthesis across target dataframes.
- Custom Primitive Definitions - Allows users to define custom primitives to extend the automated feature engineering process with specific logic.
- Natural Language Feature Extraction - Transforms unstructured text into structured numeric features using specialized natural language processing primitives.
- Natural Language Processing - Provides specialized natural language processing primitives to extract structured features from text.
- Text Feature Extraction - Transforms unstructured text into structured features using specialized primitives for quantitative analysis.
- Distributed Computing Engines - Offloads heavy feature computation to multiple cores or clusters using distributed computing engines.
- Distributed Data Processing Frameworks - Integrates with distributed computing frameworks to maintain performance when processing large volumes of data.
- Large-Scale Data Computation - Processes massive volumes of data by distributing feature computation across distributed clusters.
- Large-Scale Feature Transformations - Calculates engineered features across multiple CPU cores or clusters to process large-scale datasets.
- Distributed Computing - Distributes the recursive feature synthesis process across multiple cores or clusters for efficient large-scale processing.
- Computational Graphs - Uses computational graphs to define and optimize the sequence of feature engineering transformations.
- Feature Engineering - Automated generation of features from relational datasets.
- General Machine Learning - Automated feature engineering for relational and transactional data.
- Feature Engineering - Automated feature engineering for relational data.
