Momentum Study Based on Principal Component Analysis and GCPN-R Predictive Models

To investigate the effect of athletes' momentum on the performance of sports competitions, this paper proposes the GCPN-R prediction model, which is an improved GCN model based on the RNN and PCA methods for predicting the results and fluctuations at each point in the competition. First, we defined the components of momentum and extracted the principal components of each component factor using principal component analysis (PCA) to obtain the momentum score reflecting the athlete's performance. Second, we propose the GCPN-R prediction model, which is an improved GCN model based on the RNN and PCA methods for predicting the results and fluctuations at various points in the competition. The model takes the players' performances as input and outputs the predicted outcomes and fluctuations of all the participants at each point in a pairwise tournament. The experimental results from the validation set show that the model has more than 72.12% in predicting the outcome and fluctuation of the match. The research not only refines outcome predictions but also offers deeper insights into the impact of dynamic performance factors on match results.