A Decision Tree Model for Classification of Winning and Losing Probability from Fitness and Locomotor Parameters in Youth Badminton

In this chapter, we developed a decision tree (DT) classification model to predict competitive outcomes among youth badminton players based on a set of several fitness and motor performance parameters. A total of 67 youth badminton players drawn from different youth badminton programs across Malaysia participated in the study. The DT model was trained on 70% of the data with stratified sampling to preserve class balance. Feature selection and hyperparameter tuning were applied to enhance generalisation and reduce overfitting. The final model achieved strong performance, with training accuracy of 0.89 and AUC of 0.93, and test accuracy of 0.86 and AUC of 0.88. Balanced precision and recall of 0.91 on the test set confirmed the model’s robustness. The tree architecture revealed total points, velocity, maturity offset, and flexibility as key predictors of winning outcomes. These findings demonstrate the utility of interpretable machine learning models in projecting the match outcome winning probability in youth badminton performance, which is non-trivial for talent identification and training strategies.