Ultra-marginal Feature Importance

Joseph Janssen, Vincent Guan

Keywords: Feature Importance, AI Fairness, Interpretable Machine Learning, Removing Dependencies

TL;DR: We introduce a new feature importance framework for scientists who want to quantify the strengths of relationships in data.

Abstract: Scientists frequently prioritize learning from data rather than training the best possible model; however, research in machine learning often prioritizes the latter. Marginal contribution feature importance (MCI) was developed to break this trend by providing a useful framework for quantifying the relationships in data in an interpretable fashion. In this work, we aim to improve upon the theoretical properties, performance, and runtime of MCI by introducing ultra-marginal feature importance (UMFI), which uses preprocessing methods from the AI fairness literature to remove dependencies in the feature set prior to measuring predictive power. We show on real and simulated data that UMFI performs better than MCI, especially in the presence of correlated interactions and unrelated features, while partially learning the structure of the causal graph and reducing the exponential runtime of MCI to super-linear.

Community Implementations: [ 2 code implementations](https://www.catalyzex.com/paper/ultra-marginal-feature-importance/code)