Tennis Game Swing Prediction Model Based on ID3 Algorithm and Logistic Regression Model

Abstract

Tennis, as one of the most popular sports worldwide, includes exciting matches between players. In some matches, score advantage can swing constantly from one side to another. These score swings may impact players’ performance and the result. To accurately predict the swings during a match between players, the article aims at building a score swing prediction model based on ID3 algorithm and logistic regression algorithm. The article will focus on analyzing tennis players’ behaviors and performance during a match. The preparation is to form a decision tree to evaluate diverse events’ impact on players’ mental strength, using ID3 algorithm that can analyze desired indicators with relevant data. By using the results from the decision tree evaluation, indicators with higher priority will be analyzed by the logistic regression model. The following Omnibus test and Hosmer-Lemeshow tests will estimate significance value and prediction accuracy. The results can show the prediction of scores swings in a match and indicate the factors that impact the match more. The study can provide tennis players with theoretical advice on their training activities, which may improve the ability of reversing the match and being consistent under pressure effectively.