Research on Olympic Performance Prediction System Based on Multi-source Data Fusion

Abstract

This paper proposes an Olympic performance prediction system based on multi-source data fusion. First, through data cleaning and feature engineering, we constructed a dataset containing features such as athlete scale, number of events, and host country identification, and introduced an advantage score to quantify national competitiveness in different events. Random forest algorithm was used to predict the medal distribution of the 2028 Olympics, and a linear regression model was employed to quantitatively analyze the impact of elite coaches on medal acquisition. Experimental results show that the prediction system achieved an R² value of 0.716 on the test set, outperforming other machine learning models. Meanwhile, the study found that the coaching factor has a significant contribution to improving competitive performance, providing important reference for countries to optimize their coaching configuration.