Beyond Single-Feature Importance with ICECREAM

Michael Oesterle, Patrick Blöbaum, Atalanti A. Mastakouri, Elke Kirschbaum
Proceedings of the Fourth Conference on Causal Learning and Reasoning, PMLR 275:359-389, 2025.

Abstract

Which set of features was responsible for a certain output of a machine learning model? Which components caused the failure of a cloud computing application? These are just two examples of questions we are addressing in this work by Identifying Coalition-based Explanations for Common and Rare Events in Any Model (ICECREAM). Specifically, we propose an information-theoretic quantitative measure for the influence of a coalition of variables on the distribution of a target variable. This allows us to identify which set of factors is essential to obtain a certain outcome, as opposed to well-established explainability and causal contribution analysis methods that rank individual factors. In experiments with synthetic and real-world data, we show that ICECREAM outperforms state-of-the-art methods for explainability and root cause analysis, and achieves impressive accuracy in both tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v275-oesterle25a, title = {Beyond Single-Feature Importance with ICECREAM}, author = {Oesterle, Michael and Bl\"{o}baum, Patrick and Mastakouri, Atalanti A. and Kirschbaum, Elke}, booktitle = {Proceedings of the Fourth Conference on Causal Learning and Reasoning}, pages = {359--389}, year = {2025}, editor = {Huang, Biwei and Drton, Mathias}, volume = {275}, series = {Proceedings of Machine Learning Research}, month = {07--09 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v275/main/assets/oesterle25a/oesterle25a.pdf}, url = {https://proceedings.mlr.press/v275/oesterle25a.html}, abstract = {Which set of features was responsible for a certain output of a machine learning model? Which components caused the failure of a cloud computing application? These are just two examples of questions we are addressing in this work by Identifying Coalition-based Explanations for Common and Rare Events in Any Model (ICECREAM). Specifically, we propose an information-theoretic quantitative measure for the influence of a coalition of variables on the distribution of a target variable. This allows us to identify which set of factors is essential to obtain a certain outcome, as opposed to well-established explainability and causal contribution analysis methods that rank individual factors. In experiments with synthetic and real-world data, we show that ICECREAM outperforms state-of-the-art methods for explainability and root cause analysis, and achieves impressive accuracy in both tasks.} }
Endnote
%0 Conference Paper %T Beyond Single-Feature Importance with ICECREAM %A Michael Oesterle %A Patrick Blöbaum %A Atalanti A. Mastakouri %A Elke Kirschbaum %B Proceedings of the Fourth Conference on Causal Learning and Reasoning %C Proceedings of Machine Learning Research %D 2025 %E Biwei Huang %E Mathias Drton %F pmlr-v275-oesterle25a %I PMLR %P 359--389 %U https://proceedings.mlr.press/v275/oesterle25a.html %V 275 %X Which set of features was responsible for a certain output of a machine learning model? Which components caused the failure of a cloud computing application? These are just two examples of questions we are addressing in this work by Identifying Coalition-based Explanations for Common and Rare Events in Any Model (ICECREAM). Specifically, we propose an information-theoretic quantitative measure for the influence of a coalition of variables on the distribution of a target variable. This allows us to identify which set of factors is essential to obtain a certain outcome, as opposed to well-established explainability and causal contribution analysis methods that rank individual factors. In experiments with synthetic and real-world data, we show that ICECREAM outperforms state-of-the-art methods for explainability and root cause analysis, and achieves impressive accuracy in both tasks.
APA
Oesterle, M., Blöbaum, P., Mastakouri, A.A. & Kirschbaum, E.. (2025). Beyond Single-Feature Importance with ICECREAM. Proceedings of the Fourth Conference on Causal Learning and Reasoning, in Proceedings of Machine Learning Research 275:359-389 Available from https://proceedings.mlr.press/v275/oesterle25a.html.

Related Material