Enhancing Tennis Match Outcome Predictions Through the Integration of XGBOOST and NSGA-II Algorithms

The advance warning of the winner of tennis plays a vital role in the process of competitive games. This paper predicts the results of tennis games by combining machine learning (ML) technology, especially the eXtreme Gradient Boosting (XGBOOST) model and the non-dominated sorting genetic algorithm II (NSGA-II). The research focuses on factors such as the speed of the ball, the direction of the serve, the proximity to the boundary line when serving, and the distance of the player when hitting the ball. Through extensive data analysis, this study achieved complex predictions of the results of the game. The results showed that compared with XGBOOST, the model optimizes the prediction accuracy to 0.93 (Accuracy), 0.98 (F1), 0.97 (AUC-ROC), 0.08 (Brier score). This type of model is very valuable for predicting the results of competitive sports and studying the rules and reasons for winning.