Research on Match Competition Prediction Algorithm Based on LightGBM

With the advent of the era of big data, artificial intelligence and machine learning have developed rapidly. At present, there are many competitive competition data, which can objectively predict the level of players in the future through data analysis. Firstly, based on the data of Wimbledon men's singles in 2023, this paper preprocesses and constructs the index system, and carries out the consistent processing and correlation analysis of the indicators. Among them, player 1 serve depth, player 1 serve speed, player 2 serve speed, player 1 break success rate, player 2 break success rate are the most positively correlated, and finally 5 indicators are determined. Then, the LightGBM prediction algorithm is established to train the 185 games data of five matches, and the results of 42 games in the next game are predicted. The prediction accuracy is 83.72%. Compared with the random forest algorithm and ridge regression, the prediction algorithm has the highest accuracy. Finally, the LightGBM prediction competition fluctuation model was applied to other players, and it was found that the prediction accuracy of the model was 86.21 %. The prediction results show that the generalization effect of LightGBM algorithm is better, which provides a scientific basis for the level prediction of players in competitive competitions.