Classification of Movement Intensity Profiles in Youth Badminton Players Using Bio-Physiological Parameters: A Ridge-Kernel SVM Technique

This study aimed to classify youth badminton players into movement intensity profiles using a ridge-kernel Support Vector Machine (SVM) model, integrating sensor-derived angular velocity with physiological and maturity-re- lated parameters. A total of 67 senior youth players (mean age = 14.13 ± 1.67 years; badminton experience = 5.55 ± 2.05 years) were assessed for 17 anthropo- metric, fitness, and maturity-related parameters. Total angular velocity was com- puted from tri-axial inertial sensors and used to cluster players into Higher Move- ment Intensity (HMI) and Lower Movement Intensity (LMI) groups. A ridge- based SVM model with radial basis function kernel was developed and evaluated using a 70:30 train–test split and 3-fold cross-validation. Classification metrics included accuracy, AUC, precision, recall, F1-score, Cohen’s kappa, and MCC. Permutation importance analyses were performed to identify the most contrib- uting parameters towards model performance efficacy. The model achieved strong performance on the test set (Accuracy = 0.86, AUC = 0.88). Total angular velocity was the most influential feature, followed by plank duration, agility, aer- obic endurance, and maturity offset. The model showed high specificity but mod- erate sensitivity to HMI classification. A ridge-kernel SVM integrating move- ment, fitness, and maturation data effectively classified players’ intensity profiles. The findings of this investigation could assist in establishing an individualized training regimen in youth badminton development programs.