Research On the Performance Evaluation of Tennis Players Based on Stacking

Abstract

Evaluating the performance of tennis players in the competition not only improves the scientific and targeted training, but also realizes the accurate prediction of the match results, and promotes the fairness and transparency of competitive sports. At the same time, it enhances the spectator experience of the game, promotes the innovation and development of sports science and technology, and is of great significance to the individual growth of athletes and the overall progress of the sports industry. In order to evaluate the performance of athletes in the competition in real time, this paper establishes the PSPPM model, which combines the dynamic change and trend of time series, visualizes the competition process, and predicts the athlete's score in real time. The model uses a stacking approach, including logistic regression and XGBoost models. Stacking, as a multi-model fusion method, combines the interpretability of logistic regression, the visualization results, the feature importance assessment capability of XGBoost, and the ability to capture dynamic changes and trends in time series. This enhances the overall performance of the model, demonstrating better generalization capabilities when dealing with complex problems.