Medal Prediction Model Based on ARIMA-XGBoost Model

The Olympic Games, a premier global sporting event, mirror a nation's prowess in international sports. Accurate predictions of the medal table can aid countries in strategizing to enhance their performance. Leveraging historical medal and athlete data, we devised an ARIMA-XGBoost model to project each country's medal tally. Initially, we preprocessed and integrated the data to form a multi-dimensional feature system encompassing athlete, host, event, and medal information. Subsequently, we constructed an ARIMA-XGBoost hybrid model to predict medal counts based on these features. Furthermore, we analyzed the correlation between various factors and medal outcomes through data analysis. By conducted a sensitivity analysis of key model parameters, such as tree depth, learning rate, and sampling ratio, and we evaluated the model's strengths and weaknesses, comprehensively assess its performance.