Research on Momentum Prediction Model Based on AHP and Rusboost Algorithm

This paper focuses on the problem of momentum quantification in tennis matches, and systematically analyzes the factors that affect the momentum of matches by using momentum quantification (MQ) model and swing prediction (SP) model. First, by selecting the indicators directly related to the match (such as winning points, losing points, service points, etc.) and additional influencing factors (such as opponents' service points, winning balls, etc.), and constructing model expressions, MQ model is established. Based on analytic Hierarchy Process (AHP), the model results are solved and analyzed, thereby quantifying the momentum and predicting the dynamic change of competition. Secondly, by precisely defining turning point and selecting innovation index, the SP swing prediction model is established by RUSBoost algorithm, and the model is evaluated and compared by confusion matrix and ROC curve analysis. Finally, the validity of SP model in forecasting accuracy is verified through testing and extension, which provides a powerful tool and method for momentum quantification and swing prediction.