Research on Momentum Evaluation Based on Entropy Weighting Method and Logistic Regression

Logistic regression is a supervised learning algorithm widely used in binary classification scenarios. It can predict the probability of time occurrence with probability. In this paper, to expand its application scenario, it is applied to the prediction of momentum in tennis matches. Firstly, the factors affecting the match are classified into two categories, positive and negative, totaling twenty-seven. After that, the impact of weights is determined by the entropy weighting method through an innovative momentum assessment model, which visualizes the momentum changes of the participants. Running tests and Spearman's correlation analyses show that momentum changes are significantly correlated with race outcomes, which helps predict winners and losers. Finally, this paper defines match and momentum turning points and verifies their accuracy through logistic regression analyses, providing decision support for coaches and athletes. The prediction accuracy was further improved by including the ages of the top 30 tennis players in the world. Various evaluation metrics indicate that the model has high accuracy and generalizability. The research well extends the application of logistic regression algorithm also provides a reference for the prediction and evaluation of similar invisible factors.