Evaluating a computer based skills acquisition trainer to classify badminton players.

Area of Science:

• Sports Science
• Biomechanical Analysis
• Machine Learning in Sports

Background:

• The Skill Acquisition and Training System (SATB) program requires statistical validation for its effectiveness in classifying badminton players.
• Objective assessment of player skill levels is crucial for tailored training programs in sports like badminton.

Purpose of the Study:

• To compare the statistical predictive accuracy of neural networks and discriminant function analysis (DFA) within the SATB program.
• To evaluate the SATB program's ability to accurately classify badminton players across different skill levels (advanced, intermediate, beginner).

Main Methods:

• Utilized neural networks and discriminant function analysis to analyze data from 41 badminton players.
• Classified participants into advanced, intermediate, and beginner skill groups based on SATB program assessments.

Main Results:

• Neural networks demonstrated superior effectiveness in predicting group membership compared to DFA.
• Neural networks exhibited higher predictive validity, indicating greater accuracy in skill level classification.
• Predicting shot type was more successful than predicting location placement, highlighting trajectory assessment importance.

Conclusions:

• Neural networks offer a more robust statistical tool for skill level classification in badminton compared to DFA.
• The findings support the integration of visual-based training methods and trajectory analysis in badminton athlete development.
• The SATB program's accuracy is validated, providing valuable insights for coaches and trainers to optimize training strategies.