Feature Importance Metrics in the Presence of Missing Data

Henrik Von Kleist, Joshua Wendland, Ilya Shpitser, Carsten Marr
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:61769-61789, 2025.

Abstract

Feature importance metrics are critical for interpreting machine learning models and understanding the relevance of individual features. However, real-world data often exhibit missingness, thereby complicating how feature importance should be evaluated. We introduce the distinction between two evaluation frameworks under missing data: (1) feature importance under the full data, as if every feature had been fully measured, and (2) feature importance under the observed data, where missingness is governed by the current measurement policy. While the full data perspective offers insights into the data generating process, it often relies on unrealistic assumptions and cannot guide decisions when missingness persists at model deployment. Since neither framework directly informs improvements in data collection, we additionally introduce the feature measurement importance gradient (FMIG), a novel, model-agnostic metric that identifies features that should be measured more frequently to enhance predictive performance. Using synthetic data, we illustrate key differences between these metrics and the risks of conflating them.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-von-kleist25a, title = {Feature Importance Metrics in the Presence of Missing Data}, author = {Von Kleist, Henrik and Wendland, Joshua and Shpitser, Ilya and Marr, Carsten}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {61769--61789}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/von-kleist25a/von-kleist25a.pdf}, url = {https://proceedings.mlr.press/v267/von-kleist25a.html}, abstract = {Feature importance metrics are critical for interpreting machine learning models and understanding the relevance of individual features. However, real-world data often exhibit missingness, thereby complicating how feature importance should be evaluated. We introduce the distinction between two evaluation frameworks under missing data: (1) feature importance under the full data, as if every feature had been fully measured, and (2) feature importance under the observed data, where missingness is governed by the current measurement policy. While the full data perspective offers insights into the data generating process, it often relies on unrealistic assumptions and cannot guide decisions when missingness persists at model deployment. Since neither framework directly informs improvements in data collection, we additionally introduce the feature measurement importance gradient (FMIG), a novel, model-agnostic metric that identifies features that should be measured more frequently to enhance predictive performance. Using synthetic data, we illustrate key differences between these metrics and the risks of conflating them.} }
Endnote
%0 Conference Paper %T Feature Importance Metrics in the Presence of Missing Data %A Henrik Von Kleist %A Joshua Wendland %A Ilya Shpitser %A Carsten Marr %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-von-kleist25a %I PMLR %P 61769--61789 %U https://proceedings.mlr.press/v267/von-kleist25a.html %V 267 %X Feature importance metrics are critical for interpreting machine learning models and understanding the relevance of individual features. However, real-world data often exhibit missingness, thereby complicating how feature importance should be evaluated. We introduce the distinction between two evaluation frameworks under missing data: (1) feature importance under the full data, as if every feature had been fully measured, and (2) feature importance under the observed data, where missingness is governed by the current measurement policy. While the full data perspective offers insights into the data generating process, it often relies on unrealistic assumptions and cannot guide decisions when missingness persists at model deployment. Since neither framework directly informs improvements in data collection, we additionally introduce the feature measurement importance gradient (FMIG), a novel, model-agnostic metric that identifies features that should be measured more frequently to enhance predictive performance. Using synthetic data, we illustrate key differences between these metrics and the risks of conflating them.
APA
Von Kleist, H., Wendland, J., Shpitser, I. & Marr, C.. (2025). Feature Importance Metrics in the Presence of Missing Data. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:61769-61789 Available from https://proceedings.mlr.press/v267/von-kleist25a.html.

Related Material