Interpretable Machine Learning using SHAP — theory and applications

Interpretable Machine Learning using SHAP — theory and applications

SHAP is an increasingly popular method used for interpretable machine learning. This article breaks down the theory of Shapley Additive Values and illustrates with a few practical examples.

Introduction

Complex machine learning algorithms such as the XGBoost have become increasingly popular for prediction problems. Traditionally, there has been a trade-off between interpretation and accuracy, and simple models such as the linear regression are sometimes preferred …

applications data science learning machine machine learning shap statistics theory