Crunching the Numbers: How Machine Learning Predicts South Korean Badminton Champ's Moves

In the world of badminton, South Korea's An Se-young has been dominating the scene with an impressive win rate. Using machine learning, researchers have analyzed her game, aiming to predict how she'll score and what tactics she'll use. The study looked at her play in the 2023 BWF Olympic points cycle, where she clocked an 89. 5% win rate. The team used a "three-stage" data classification to break down her techniques and tactics.

They didn't just rely on statistics, though. They also watched 10 of her matches, each with 21 points, where she faced top-ranked players. This hands-on approach helped refine their data. Using algorithms like 'Decision tree', 'Random forest', 'XGBoost', 'Support vector', and 'K-proximity', they built a model to guess her moves. The support vector machine with the RBF function kernel was the star, hitting an accuracy of 87. 5%. An's style is all about pouncing on her opponents' mistakes. She doesn't go for overwhelming pressure. Instead, she capitalizes on any slip-ups with quick attacks like kills or dives. This strategy often turns the tide in her favor within just a few strikes. The study found that this calculated approach is both effective and consistent in predicting her gameplay.

Actions