Familywise Feature Importance Stability in Chemical and Materials Machine Learning
Journal of Chemical Information and Modeling · 27.07.2026 · 66 (14) · 7996-8007
DOI: 10.1021/acs.jcim.6c00893 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Chemical Information and Modeling, online erschienen am 16.07.2026. Verfasst von Ruilin Lai und 4 weiteren Autor:innen. Beteiligte Einrichtung: College of Computer Science, Beijing Information Science and Technology University, Beijing100101, P. R. China. Hauptthemen laut MEDLINE: Materials Science, Machine Learning. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 44 Literaturangaben. Gefördert von: Natural Science Foundation of Beijing Municipality, National Natural Science Foundation of China.
Abstract (Originalsprache)
While feature importance analysis in chemical and materials machine learning can often be sensitive to both the predictive model and the attribution rule, the robustness of these rankings is rarely quantified before they are used to gain physical insights. Here, we compare 26 feature importance pipelines spanning data-driven, model-based, and formula-based analyses on a metal–support interaction data set anchored by an explicit SISSO equation, and we examine whether the same qualitative behavior recurs in high-entropy-alloy and halide perovskite data sets. Across the three benchmarks, we observe high intrafamily agreement but substantial interfamily variance. While a small subset of features remains stable across multiple families, several midranked features are highly family dependent, with their apparent importance shifting according to the underlying modeling assumptions. To ensure robust interpretability, we recommend that feature importance be reported by method family or correlation-based clusters, supplemented by resampling intervals.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- College of Computer Science, Beijing Information Science and Technology University, Beijing100101, P. R. China.
- Beijing Advanced Innovation Center for Materials Genome Engineering, Beijing Information Science and Technology University, Beijing100101, P. R. China.
- UCLouvain , Chemin des Étoiles 8, Louvain-la-Neuve1348, Belgium.
- Department of Chemical Physics, School of Chemistry and Materials Science, University of Science and Technology of China, Anhui230026, P. R. China.
- Key Laboratory of Precision and Intelligent Chemistry, University of Science and Technology of China, Anhui230026, P. R. China.
ORCID-Profile: Xiaotong Liu, Gian-Marco Rignanese, Wei-Xue Li
Förderung
- Natural Science Foundation of Beijing Municipality
- National Natural Science Foundation of China
Zitieren
APA
Lai, R., Liu, X., Wang, Y., Rignanese, G. M., & Li, W. X. (2026). Familywise Feature Importance Stability in Chemical and Materials Machine Learning. Journal of Chemical Information and Modeling, 66(14), 7996-8007. https://doi.org/10.1021/acs.jcim.6c00893
Vancouver
Lai R, Liu X, Wang Y, Rignanese GM, Li WX. Familywise Feature Importance Stability in Chemical and Materials Machine Learning. Journal of Chemical Information and Modeling. 2026;66(14):7996-8007. doi:10.1021/acs.jcim.6c00893
BibTeX
@article{Lai2026,
author = {Lai, Ruilin and Liu, Xiaotong and Wang, Yuhang and Rignanese, Gian-Marco and Li, Wei-Xue},
title = {Familywise Feature Importance Stability in Chemical and Materials Machine Learning},
journal = {Journal of Chemical Information and Modeling},
year = {2026},
volume = {66},
number = {14},
pages = {7996--8007},
doi = {10.1021/acs.jcim.6c00893}
}
EUVIMED