ShapShift: Explaining Model Prediction Shifts with Subgroup Conditional Shapley Values

2026-04-13

AI Summary

This work addresses the challenge of prediction mean shifts caused by changes in input data distributions, which can severely disrupt downstream decision-making and necessitate interpretable attribution mechanisms. The authors propose a novel approach based on subgroup-conditional Shapley values, leveraging decision tree structures to define semantically meaningful subgroups. The method attributes prediction shifts to changes in the conditional probabilities of these subgroups and extends naturally to tree ensembles and model-agnostic settings. Its core innovation lies in the first-time integration of Shapley values with decision tree–derived subgroup conditional probabilities, enhanced by representative tree selection, residual correction, and a surrogate tree fitted with a new objective function. This framework delivers efficient, faithful, and nearly complete attributions, significantly improving the interpretability and monitoring of prediction shifts across diverse models in dynamic environments.

Technology Category

• Machine Learning: Ensemble Methods
• Reasoning under Uncertainty: Probabilistic Programming
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

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.

Problem

• prediction shift
• input distribution shift
• model monitoring
• conditional probability
• interpretable subgroups

Innovation

• Shapley values
• prediction shift
• subgroup attribution
• surrogate trees
• model-agnostic explanation