Author(s)

Yanxi Fan, Guowei Zhou

Corresponding Author

Yanxi Fan

Abstract

This study integrates machine learning with multi-criteria decision analysis to develop a data-driven model for Olympic medal prediction and sports strategy planning. We employed an XGBoost-Bootstrap ensemble model (R²=0.9579, RMSE=0.2131) with eight critical features to predict medal counts. Our model forecasts that the United States will maintain its medal table dominance in 2028 (123-138 medals, 95% confidence interval), while countries like Germany will improve and Japan and France may decline. Using XGBoost classification, we predict Bangladesh, Kiribati, and Albania to win their first Olympic medals. Through Spearman's correlation and K-means++ analysis, we identified athletics and swimming as strategic priorities for most countries. The AHP-CRITIC model quantifies the "great coach" effect, recommending optimal sports investments for Brazil (basketball), South Africa (athletics), and Denmark (shooting). Our analysis reveals two key trends in Olympic medal distribution: the "Great Power Effect" and "Home Advantage Effect," providing valuable insights for national Olympic committees in resource allocation and strategic planning. The practical applications of this research extend beyond medal prediction to strategic resource allocation, helping identify high-potential sports for specific countries and quantifying coaching investment impacts for more efficient distribution of limited resources.

Keywords

Medal Prediction, XGBoost, K-means++, AHP-CRITIC, Home Advantage