Optimizing Prediction Accuracy in the English Premier League: A Comparative Analysis of Machine Learning Models for Forecasting Match Outcomes

This study presents a comparative analysis of machine learning models on their ability to determine match outcomes in the English Premier League (EPL), focusing on optimizing prediction accuracy. The research leverages a variety of models, including logistic regression, decision trees, random forests, gradient boosting machines, support vector machines, k-nearest neighbors, and extreme gradient boosting, to predict the outcomes of soccer matches in the EPL. Utilizing a comprehensive dataset from Kaggle, the study uses the Sport Result Prediction CRISP-DM framework for data preparation and model evaluation, comparing the accuracy, precision, recall, F1-score, ROC-AUC score, and confusion matrices of each model used in the study. The findings reveal that ensemble methods, notably Random Forest and Extreme Gradient Boosting, outperform other models in accuracy, highlighting their potential in sports analytics. This research contributes to the field of sports analytics by demonstrating the effectiveness of machine learning in sports outcome prediction, while also identifying the challenges and complexities inherent in predicting the outcomes of EPL matches. This research not only highlights the significance of ensemble learning techniques in handling sports data complexities but also opens avenues for future exploration into advanced machine learning and deep learning approaches for enhancing predictive accuracy in sports analytics.