Predicting Olympic Medal Counts: A Random Forest Regression with Monte Carlo Simulation Model

At every Olympic Games, the medal tally is often the focal point of global attention. In this paper, we take a different approach by leveraging regression analysis on historical data features to predict medal counts. We built a feature-enhanced random forest regression model incorporating Monte Carlo simulation, and Monte Carlo simulation is used to provide uncertainty intervals for the prediction. MSE and $R 2$ of gold medals prediction model is 5.08 and 0.856, and total medals prediction model is 39.85 and 0.842, which indicates our model fits well. We analyze the feature importance in the random forest model to get our original insights as well as recommendations to the Olympic Committee. Additionally, we performed sensitivity analysis to explore the impact of input variations on the model. One of the feature variables went up and down by $10 \%$ when the data went up and down by $13 \%$, reflecting excellent model sensitivity.