OLYMPIC MEDAL PREDICTION AND ANALYSIS BASED ON LSTM AND TOPSIS MODELS

In the context of the increasingly fierce global sports competition, accurately predicting Olympic medal outcomes and optimizing the allocation of sports resources have become crucial concerns for national sports committees and related organizations. This study tackles these challenges through the use of advanced modeling techniques. A Long Short-Term Memory (LSTM) model was constructed using historical data from the Summer Olympics (1896-2024), encompassing medal counts, participating events, and national indicators such as population and GDP. The model takes into account the time-dependence of historical performance, advantages in sports infrastructure, and the benefits of being the host country. The results predict that the United States, China, and France will demonstrate strong medal competitiveness at the 2028 Los Angeles Olympics, with potential breakthroughs from emerging nations. Moreover, a decision tree model was employed to examine the influence of "great coaches" on medal results. By examining coach mobility, athlete performance data, and changes in medal counts, the study revealed that transnational coach mobility significantly influences medal distribution. Notable coaches like Lang Ping and Bela Karolyi have enhanced the competitiveness of volleyball and gymnastics, respectively. The findings suggest that recruiting top-tier coaches can increase medal counts and elevate international sports performance. This research provides valuable strategies for optimizing sports resource allocation and enhancing global competitiveness.