Examples#
This section contains end-to-end examples demonstrating how to generate optimized rulesets using Iguanas.
Overview#
Each example notebook demonstrates a complete ML workflow using Iguanas rule generation, alone or with and feature engineering done with the gators package
Titanic Survival Prediction#
Binary classification using advanced feature engineering including string parsing, mathematical features, and rare category encoding.
Without feature engineering, the best F1 score from rules alone is 0.640:
Using a sklearn API wrapper for Iguanas, the best F1 score from rules alone is 0.642:
With feature engineering, the best F1 score from rules done on engineered features is 0.78: