Machine Learning for Optimal Player Substitutions in Soccer

Machine Learning for Optimal Player Substitutions in Soccer

---

연구 분야: Artificial Intelligence

---

---

학회: SN Computer Science

---

초록

This research examines the application of ML to predict optimal times for player substitutions in football, aiming to improve game outcomes through strategic decisions. We analyzed a dataset from Kaggle, containing 51,738 substitution instances across 9074 games from five top European leagues over six seasons. The study assessed various Machine Learning models, including Logistic Regression (LR), Decision Trees (DT), K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Multinomial Naïve Bayes (MNB), Random Forest (RF), and eXtreme Gradient Boosting (XGBoost), using an 80-20 data split for model testing. The findings reveal that Support Vector Machines and Multinomial Naïve Bayes were the most effective, each achieving over 79% accuracy and the highest F1-score at 89%. eXtreme Gradient Boosting displayed the highest precision at 81%, while Decision Trees had the lowest accuracy at 67%. In terms of efficiency, Multinomial Naïve Bayes was the fastest model to train, whereas Support Vector Machines required the most time. These results underscore the potential of Machine Learning to provide a tactical edge by facilitating more informed substitution decisions during football games.

---