Lasso Ridge based mostly XGBoost and Deep_LSTM Assist Tennis Gamers Carry out higher
Authors: Wankang Zhai, Yuhan Wang
Summary: Understanding the dynamics of momentum and sport fluctuation in tennis matches is cru-cial for predicting match outcomes and enhancing participant efficiency. On this research, we current a complete evaluation of those elements utilizing a dataset from the 2023 Wimbledon remaining. Ini-tially, we develop a sliding-window-based scoring mannequin to evaluate participant efficiency, ac-counting for the affect of serving dominance by way of a serve decay issue. Moreover, we introduce a novel strategy, Lasso-Ridge-based XGBoost, to quantify momentum results, lev-eraging the predictive energy of XGBoost whereas mitigating overfitting by way of regularization. By means of experimentation, we obtain an accuracy of 94% in predicting match outcomes, iden-tifying key elements influencing successful charges. Subsequently, we suggest a Spinoff of the successful price algorithm to quantify sport fluctuation, using an LSTM_Deep mannequin to pre-dict fluctuation scores. Our mannequin successfully captures temporal correlations in momentum fea-tures, yielding imply squared errors starting from 0.036 to 0.064. Moreover, we discover me-ta-learning utilizing MAML to switch our mannequin to foretell outcomes in ping-pong matches, although outcomes point out a comparative efficiency decline. Our findings present useful in-sights into momentum dynamics and sport fluctuation, providing implications for sports activities analytics and participant coaching methods
