Ibrahim Cosar

Predicting Soccer Match Outcomes with ML

• Web Scraping: Scraped English Premier League (EPL) data with BeautifulSoup using class descriptors from industry websites like Sofascore and collected it in csv format.
• Feature Engineering: Processed and engineered predictive features such as form, xG, home/away performance, and goal distributions to enhance model accuracy.
• Machine Learning Approach: Applied Tree-based methods (Random Forest) to classify match results and predict probabilities based on different features.
• Entropy & Information Gain: Optimized decision trees using entropy-based splitting criteria to maximize predictive power and reduce overfitting.