OLYMPIC MEDAL PREDICTION BASED ON SEQ2SEQ MODEL AND TPE OPTIMIZATION

This paper refers to provide scientific basis for the strategic planning of sports in various countries through a high-precision Olympic medal prediction model. In this study, firstly, a comprehensive preprocessing of the raw data was carried out and the missing values were filled in using the preprocessing of (BP) neural network, secondly, a sequence-to-sequence (Seq2Seq) based model was constructed to predict the number of Olympic medals, and the model hyperparameters were optimized by using tree-structured TPE (Tree-structured Bayesian Optimization), and the uncertainties in the prediction results were quantified by using employing Monte Carlo simulation and confidence intervals The results show that the optimized model has a good performance in the testing of the model. The results showed that the coefficient of determination of the optimized model improved from 0.827 to 0.875 on the test set and then predicted the number of medals for each country in the 2028 Summer Olympics in Los Angeles. The study shows that the model can effectively predict the distribution of Olympic medals to provide a valuable reference for national sports events and sports development.