Prediction for Olympic Medal Tables by Machine Learning
Jianzhang Li, Yueran Zhang · International Journal of Biology and Life Sciences · 2025
The goal of this article is to predict the gold and overall medal rankings for 2028. ARIMA is used to analyze cyclical fluctuations, followed by XGBoost optimization for time-series predictions. The model forecasts gold medal standings and visualizes the top 20 countries. Further, this paper pay attention to predicting the likelihood of countries without Olympic gold medals winning their first in 2028. Our study create binary classification labels and countries that have never won gold are identified and relevant features are extracted. SVM is employed for classification, with an AUC score close to 1, indicating high accuracy. Eventually, the relationship between event selection and medal counts is analyzed. The data is preprocessed and event types are converted into categorical variables. Random Forest regression is used, revealing that host country event selection affects medal performance but has minimal impact on overall rankings. The model’s performance is validated using MAE, MSE, and other metrics.