Enhancing Premier League Match Outcome Prediction Using Support Vector Machine with Ensemble Techniques: A Comparative Study on Bagging and Boosting

Predicting football match outcomes is a significant challenge in sports analytics, requiring models that are both accurate and resilient. This study evaluates the effectiveness of ensemble techniques, specifically Bagging and Boosting, in enhancing the performance of Support Vector Machine (SVM) models for predicting match outcomes in the English Premier League. The dataset comprises detailed match statistics from 1,520 matches across multiple seasons, including features such as team performance, player statistics, and match outcomes. Four models were examined: baseline SVM, SVM with Bagging, SVM with Boosting, and a combined SVM + Bagging + Boosting approach. Evaluation metrics include accuracy, recall, precision, F1 score, and ROC-AUC, providing a comprehensive assessment of each model's performance. Experimental results indicate that ensemble methods substantially improve model accuracy and stability, with the SVM + Bagging + Boosting combination achieving perfect scores in accuracy, recall, precision, and F1 score, alongside an ROC-AUC value of 0.88. However, this model's slightly reduced ROC-AUC compared to others and its high computational cost highlight potential risks of overfitting and the need for significant resources. These findings underscore the practical potential of combining Bagging and Boosting with SVM for robust and accurate predictions. Limitations include the dataset's focus on a single league and the high resource requirements for ensemble methods. Future research could expand this approach to other sports and leagues, improve computational efficiency, and explore real-time predictive applications