Research on Olympic medal prediction based on GA-BP and logistic regression model [version 2; peer review: 1 approved with reservations]

• Published in: F1000 research Vol. 14; p. 245
• Main Authors: Zhao, Sanglin, Cao, Jikang, Steve, Jackon
• Format: Journal Article
• Language: English
• Published: 2025
• Subjects: Logistic regression; Virtual control group; Genetic algorithm; Coach effect; Olympic medals
• ISSN: 2046-1402; 2046-1402
• DOI: 10.12688/f1000research.161865.2

Background Predicting the number and distribution of Olympic medals in the future has become a hot topic, but predicting the number of Olympic medals is not easy and requires comprehensive consideration of multiple factors such as historical data, athlete performance, and host country effects. Method This article uses the GA-BP algorithm model, combined with genetic algorithm (GA) and backpropagation neural network (BPNN), to optimize the weights and bias parameters of the BP neural network using the global search capability of genetic algorithm, thereby improving training efficiency and prediction performance. By estimating the number of Olympic gold medals and total medals, verifying the accuracy of the model, and predicting the medal table for the 2028 Los Angeles Olympics. Meanwhile, based on the synthetic control model, Estonia and China were selected as research subjects to construct a virtual control group and two experimental groups for analysis. Result The experimental results showed that Estonia and China won more medals with a head coach than without one. In 1992, Estonia won 1 gold medal and 2 bronze medals under the guidance of excellent coaches, indicating the significant role of head coaches in improving athletes' performance. Conclusion This study provides valuable insights for the decision-making of the Olympic Committee, revealing key factors in medal distribution, optimizing the allocation of national strategic resources, and predicting the performance of countries at future Olympic Games.