Olympic Medal Prediction Through Multidimensional Features and Enhanced XGBoost Modeling

The number of Olympic medals is often an important indicator when measuring a country's sports achievement, so predicting the number of medals each country will win in the next Olympic Games becomes an important issue. Existing prediction models rely heavily on the number of medals won in previous Olympics, ignoring important factors such as athletes' recent performances and the host country's advantages, so this study proposes a multidimensional prediction model based on the XGBoost algorithm. The model takes into account the recent performance of athletes, the host country's advantages and the influence of advantageous projects on the number of medals, and uses the grey wolf optimization algorithm hyperparameters to optimize, and finally accurately predicts the medal winners of individual countries in the 2028 Olympic Games in Los Angeles. Among them, the United States will reach 50 gold medals with the addition of the host country, while China's number of gold medals will drop to 31 due to the reduction of dominant events. In the accuracy assessment, the model proposed by this study has an MSE of 0.87 and an R² of 0.95, which indicates that the model has a very superior accuracy. The medal results predicted by this study are of great strategic significance, and play a positive role in guiding the direction of each country's preparation for the Olympic Games.