Athlete Momentum Visualization Based on Logit Regression Model and LSTM Prediction Model

Abstract

With the development of information technology, big data analytics predicts the likelihood of a player's victory by capturing the game situation, thus providing the player with a timely and appropriate strategy to win the game. In order to investigate whether the player's "momentum" has an impact on the game results, this paper extracts four indicators, namely, the player difference, the number of consecutive wins, the number of breaks of serve, and the number of successful breaks of serve based on the data of the Wimbledon Men's Tennis Championships in 2023, in order to quantify the impact of the "momentum" on the game results.  Logit regression models were then developed to find strong correlations between these metrics and the number of wins. Therefore, in order to further investigate the influence of these metrics on the game process, an LSTM prediction model was built to predict the outcome of the game. The training results of the final model show that the MSE is 0.316 and the MA is 0.387, with low prediction errors. Based on the prediction results of LSTM, this paper provides appropriate suggestions for players by analyzing the turning points of win/loss tilts and changes in players' status.