Modeling of Sports Performance Based on Nonlinear Screening Factors and Weighting to Improve Prediction Accuracy
Fan Zhang · 2018
In this paper, a weighted sports performance prediction model based on LSSVM is constructed for complex non-linear sports performance. The influencing factors of sports performance are selected and weighted by LSSVM, and the factors closely related to the predicted results are screened out, and appropriate weights are given to each factor. Firstly, the least square support vector machine (LS-SVM) was used to obtain the main influencing factors by nonlinear screening according to the principle of minimum cross-validation root mean square error. Then the main influencing factors are given different weights to reflect the extent of their impact on the results of sports performance prediction. Finally, the least squares support vector machine is used to build the optimal sports performance prediction model, and applied to the 1000-meter race performance prediction. The simulation results show that LSSVM-New improves the accuracy of sports performance prediction compared with other sports performance prediction models, and it is an effective sports performance prediction model.