An Explainable Adaptive Meta-Ensemble Framework with Hybrid Feature Selection for Accurate ODI Cricket Match Outcome Prediction
PMRP Journal of Research – Part A Computational Engineering · 2026
DOI: 10.65383/jra/2026.2.2.1 ↗Kurz gefasst
Fachartikel in der Zeitschrift PMRP Journal of Research – Part A Computational Engineering, 2026. Verfasst von Deepa S und 4 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben.
Abstract (Originalsprache)
Forecasts for One Day International (ODI) cricket matches are difficult since match outcomes are influenced by interactions between variables that vary over time, such as team strength, team form, venue quality, head-to-head records, and toss outcomes, whereas presently available methods involve static ensembles with poor explainability. In this paper, an Explainable Adaptive Meta-Ensemble Framework (EAMEF) is introduced, which utilizes dynamic engineering features unique to the domain of cricket, i.e., a Boruta-SHAP hybrid feature selection pipeline, three gradient boosting base learners (XGBoost, LightGBM, CatBoost), a softmax-weighted soft voting ensemble, and an embedded SHAP-based explanation module. Five contextual variables, such as Team Strength Index, Recent Form Index, Venue Quality Index, Head-to-Head Dominance Index, and Toss Effectiveness Index, are explicitly formulated and used in combination with the interaction and rolling-window statistics. The weights of the base learners are decided based on their cross-validated performance. Using an available ODI dataset, EAMEF secured 97.25% accuracy, 97.25% F1 score, 0.9978 ROC-AUC, and 0.945 MCC, beating the performance of Hard Voting, Soft Voting, Stacking, and Blending techniques significantly. Wilcoxon signed-rank tests proved the significance of the improvements. SHAP analysis revealed that team strength, form, home advantage, and head-to-head record were the leading factors in predictions.
Zitieren
APA
S, D., C, R. K., C, J., M, A., & A, K. (2026). An Explainable Adaptive Meta-Ensemble Framework with Hybrid Feature Selection for Accurate ODI Cricket Match Outcome Prediction. PMRP Journal of Research – Part A Computational Engineering. https://doi.org/10.65383/jra/2026.2.2.1
Vancouver
S D, C RK, C J, M A, A K. An Explainable Adaptive Meta-Ensemble Framework with Hybrid Feature Selection for Accurate ODI Cricket Match Outcome Prediction. PMRP Journal of Research – Part A Computational Engineering. 2026. doi:10.65383/jra/2026.2.2.1
BibTeX
@article{S2026,
author = {S, Deepa and C, Rukumani Khandhan and C, Jamunadevi and M, Arumugam and A, Kathirmurugan},
title = {An Explainable Adaptive Meta-Ensemble Framework with Hybrid Feature Selection for Accurate ODI Cricket Match Outcome Prediction},
journal = {PMRP Journal of Research – Part A Computational Engineering},
year = {2026},
doi = {10.65383/jra/2026.2.2.1}
}
EUVIMED