SmartTrain: An SVM and Association Rule Mining-Based Model for Predicting Athletic Performance and Optimizing Training Strategies
Zhaohui Xie, Yan Huo · Journal of Circuits Systems and Computers · 2025
In individualized sports training, predicting athletic performance and optimizing training methods are essential objectives. However, conventional approaches frequently fail to adequately capture the intricate connections between different elements, including training data, performance history and an athlete’s unique traits, leading to suboptimal recommendations that do not meet the diverse needs of athletes. To address these limitations, this study proposes a novel model called SmartTrain, which integrates the Support Vector Machine (SVM) with Association Rule Mining (ARM) for performance prediction and training strategy optimization. Specifically, ARM is employed to uncover hidden patterns and relationships within training datasets, such as the influence of training intensity and recovery time on athletic performance, while SVM provides precise classification and prediction based on these extracted patterns. By combining the interpretability of ARM and the predictive accuracy of SVM, SmartTrain delivers more accurate and actionable insights for training decision-making. Experimental results show that SmartTrain significantly outperforms traditional machine learning models, achieving an accuracy of 89.3% and an F1-score of 87.2%, which is notably higher than models like standalone SVM and decision trees. These findings underline the potential of SmartTrain to improve the quality of training recommendations and enhance the accuracy of performance forecasts. This study provides a robust framework for advancing personalized sports training programs, demonstrating how data-driven methods can support real-world applications in sports coaching and athlete development.