A Study on Tennis Momentum Prediction Based on Random Forest Algorithm

As a strategic sport, tennis plays a key role in the study of its momentum from the training of players and the decision support of coaches. Firstly, this article obtains open source data of the 2023 Wimbledon 1701 competition, quantifies momentum through whether it is a serving party in the competition and personal skills, and determines 16 indicators. Secondly, a random forest model is built to predict the momentum changes of two players in this game from multiple aspects such as accuracy and AUC. The results show that although Player 1 has shown great momentum many times on the field, Player 2 has a greater momentum, especially within the first 30 minutes. Then, the fluctuations in performance in the competition and the randomness of success were evaluated. The KW test was introduced to test the "momentum" in the game, and the result showed that p-value=0.000<a=0.05, Ho was rejected, which means that the player's performance is positively correlated with the score. Finally, the assessment of the advantages and disadvantages of the model is made and some suggestions are given to provide feedback for coaches and players.