Predicting movement intensity through anthro-fitness profiling and its association with match outcomes in youth badminton: a logistic regression-based machine approach.

Abstract

Objectives: The fast-paced nature, coupled with the dynamic movement of badminton, renders movement intensity an essential indicator for success. In youth badminton, anthropometric attributes and fitness profile varies across players, yet little is known about the association of these indicators with competitive outcomes. In this study, we aim to classify movement intensity levels and examine the relationship between intensity classification and winning probability in youth badminton players.

Methods: A total of 67 youth badminton players (mean age = 14.13 ± 1.67 years; badminton experience = 5.55 ± 2.05 years) drawn from different badminton programs in Malaysia participated in the study. Inertial measurement units were used to capture movement intensity during match play, while anthropometric, physiological and fitness evaluations were completed.

Results: K-means clustering grouped players into high and low movement intensities (HMI; LMI), respectively. The logistic regression model yielded an excellent performance in classifying the HMI and LMI with 97% accuracy. Feature importance analysis identified 11 essential predictors comprising velocity, agility, reaction time, core muscle endurance, height and weight as the most influential variables in differentiating intensity levels. Match outcome analysis revealed that HMI players had a winning probability of 68.4%, compared to 60.4% for LMI players.

Conclusion: The ability for players to sustain higher movement intensity during match play is non-trivial for ensuring success in youth badminton. The application of machine learning in athlete profiling could assist coaches in enhancing training design and workload monitoring, leading to long-term success in youth badminton.