Optimal model for predicting highest runs chase outcomes in T-20 international cricket using modern classification algorithms

This study examines the theoretical framework and practical application of machine learning models to predict high run chase outcomes in T-20 International (T-20I) cricket. Focusing on factors such as venue, toss outcomes, batting order, pitch conditions, and team rankings, the analysis investigates how these elements contribute to the probability of achieving high-scoring chases. A dataset comprising 458 T-20I matches from 2005 to 2024 was analyzed using seven machine learning models: Random Forest, Support Vector Machine, Naïve Bayes, Neural Networks, Decision Tree, XGBoost, and K-Nearest Neighbors. Model performance was systematically evaluated through metrics including accuracy, precision, recall, and Area under the Curve (AUC), with robustness confirmed via Monte Carlo simulations across variable match conditions. Results highlight Random Forest and Decision Tree models as the most accurate and reliable, consistent with theoretical expectations for ensemble and decision-based models in complex, multi-factor analyses. This research provides both theoretical and practical insights to enhance data-driven strategies, pre-match evaluations, and in-game decision-making for T-20I cricket performance analysis.Top of Form