A Comparative Study of Methods for Estimating Conditional Shapley Values and When to Use Them

TL;DR

This paper compares various methods for estimating conditional Shapley values in machine learning, evaluating their accuracy and practicality through simulations and real data, and provides guidelines for choosing the appropriate approach.

Contribution

It introduces new methods, extends existing approaches, and systematically compares different classes of methods for estimating conditional Shapley values.

Findings

Parametric methods are most accurate when data distribution is well specified.

Regression-based methods are slower to train but faster at explanation time.

Monte Carlo methods are suitable when quick explanations are needed without extensive training.

Abstract

Shapley values originated in cooperative game theory but are extensively used today as a model-agnostic explanation framework to explain predictions made by complex machine learning models in the industry and academia. There are several algorithmic approaches for computing different versions of Shapley value explanations. Here, we focus on conditional Shapley values for predictive models fitted to tabular data. Estimating precise conditional Shapley values is difficult as they require the estimation of non-trivial conditional expectations. In this article, we develop new methods, extend earlier proposed approaches, and systematize the new refined and existing methods into different method classes for comparison and evaluation. The method classes use either Monte Carlo integration or regression to model the conditional expectations. We conduct extensive simulation studies to evaluate how precisely the different method classes estimate the conditional expectations, and thereby the conditional Shapley values, for different setups. We also apply the methods to several real-world data experiments and provide recommendations for when to use the different method classes and approaches. Roughly speaking, we recommend using parametric methods when we can specify the data distribution almost correctly, as they generally produce the most accurate Shapley value explanations. When the distribution is unknown, both generative methods and regression models with a similar form as the underlying predictive model are good and stable options. Regression-based methods are often slow to train but produce the Shapley value explanations quickly once trained. The vice versa is true for Monte Carlo-based methods, making the different methods appropriate in different practical situations.