Performance and Score Analysis of Athletes Based on EWM-TOPSIS Algorithm and GSRF Prediction Model

This paper presents an analysis of athlete performance and scoring based on the EWM-TOPSIS algorithm and GSRF prediction model in the context of tennis. The EWM-TOPSIS algorithm is employed to score each point of tennis players, providing a real-time representation of the match's progress. The study reveals intense competition between players Carlos and Nicholas, both being formidable opponents. In the initial two games of the first set, Nicholas marginally leads Carlos by 0.00374 points. Furthermore, to effectively capture the dynamics of the game, a logistic regression model is developed to predict Carlos's scoring ability at each point. The model achieves an average accuracy of 92.4 % and 84.0% on training and prediction data, respectively, demonstrating high regression prediction accuracy. Additionally, a GSRF prediction model is developed to forecast Carlos's second-half performance, utilizing a grid search algorithm for optimal parameter selection in a random forest model. The model, using six metrics including winning shots, net shots, and receptions as inputs, demonstrates promising training results with an R2 of 0.7715 and MSE of 1.8210e-06, indicating a strong fitting effect.