ShapShift: Explaining Model Prediction Shifts with Subgroup Conditional Shapley Values
For practitioners monitoring ML models in production, this method offers a way to explain why average predictions change over time, but the approach is incremental as it builds on existing Shapley value and decision tree techniques.
The paper proposes ShapShift, a Shapley value method to attribute prediction shifts in machine learning models to changes in conditional probabilities of interpretable subgroups defined by decision tree structures. The method provides simple, faithful, and near-complete explanations across model classes, aiding model monitoring in dynamic environments.
Changes in input distribution can induce shifts in the average predictions of machine learning models. Such prediction shifts may impact downstream business outcomes (e.g. a bank's loan approval rate), so understanding their causes can be crucial. We propose \ours{}: a Shapley value method for attributing prediction shifts to changes in the conditional probabilities of interpretable subgroups of data, where these subgroups are defined by the structure of decision trees. We initially apply this method to single decision trees, providing exact explanations based on conditional probability changes at split nodes. Next, we extend it to tree ensembles by selecting the most explanatory tree and accounting for residual effects. Finally, we propose a model-agnostic variant using surrogate trees grown with a novel objective function, allowing application to models like neural networks. While exact computation can be intensive, approximation techniques enable practical application. We show that \ours{} provides simple, faithful, and near-complete explanations of prediction shifts across model classes, aiding model monitoring in dynamic environments.