Predictive Model Construction Based on GA And K-Means Algorithm

Abstract

In this paper, a data trend prediction framework based on genetic algorithm (GA) and K-means cluster analysis is proposed, focusing on the synergistic application of computational intelligence algorithms in complex data modeling. First, a prediction model for the total number of medals and the number of gold medals is constructed by the GA algorithm. Second, K-means cluster analysis was utilized to quantify the competitive sports strengths of the participating countries into discrete classes, which were entered into the regression equation as dummy variables to significantly improve the goodness-of-fit. In addition, the RCA index was introduced to quantify country-sports comparative advantage and to analyze the relevance of medal distribution. The model was validated for predictive accuracy by RMSE, MAPE and MAE, and the contribution of coaches was quantified based on residual analysis. The experimental results show that the GA regression model has the smallest medal prediction error, and the feature coding mechanism enhances the interpretability of the model, providing an efficient computational framework for Olympic performance prediction.