ML Feature Engineering

Purpose

Transform raw data into effective ML features through encoding, scaling, extraction, interaction, and selection. Use automated feature engineering for relational and time-series data.

Agent Protocol

Trigger

Exact user phrases: "feature engineering", "featuretools", "tsfresh", "feature selection", "feature extraction", "encoding", "scaling", "one-hot encoding", "target encoding", "feature interaction", "datetime features", "text features", "feature importance", "categorical encoding", "TF-IDF", "count vectorizer".

Input Context

Before activating, verify:

• Data sources (relational tables, CSV, time-series, text, images)
• Feature types (numeric, categorical, datetime, text, geospatial)
• Target variable (regression, classification, time-series forecasting)
• Dataset size (rows, columns, total memory usage)
• ML model type (linear models, tree-based, neural networks)
• Domain knowledge (business rules, known interactions, feature semantics)
• Infrastructure (Python environment, memory constraints, compute budget)