Explaining predictive models with mixed features using Shapley values and conditional inference trees
This work addresses the need for interpretable AI in domains like finance, though it is incremental by extending existing Shapley value methods to handle mixed dependent features.
The paper tackles the problem of explaining predictions from black-box models with mixed dependent features by proposing a method that uses Shapley values and conditional inference trees to model feature dependencies. It demonstrates that this method often outperforms current standards in simulations and shows competitive results on a real financial dataset from the FICO challenge.
It is becoming increasingly important to explain complex, black-box machine learning models. Although there is an expanding literature on this topic, Shapley values stand out as a sound method to explain predictions from any type of machine learning model. The original development of Shapley values for prediction explanation relied on the assumption that the features being described were independent. This methodology was then extended to explain dependent features with an underlying continuous distribution. In this paper, we propose a method to explain mixed (i.e. continuous, discrete, ordinal, and categorical) dependent features by modeling the dependence structure of the features using conditional inference trees. We demonstrate our proposed method against the current industry standards in various simulation studies and find that our method often outperforms the other approaches. Finally, we apply our method to a real financial data set used in the 2018 FICO Explainable Machine Learning Challenge and show how our explanations compare to the FICO challenge Recognition Award winning team.