Stress Level Classifier: Taiwanese College Table Tennis Athletes’ Electroencephalography Analysis Based on Decision Trees

Abstract

This study aims to provide a method to quantify the stress level with numerical EEG values, identify key features of brainwave and assess the level of stress for table tennis players. The data of College’s Division 1 and Division 2 are collected and analyzed by the decision tree algorithms C4.5, CART, Random Forest and Random Tree. The result of Random Forest obtains the highest accuracy rate among each algorithm, which is 79.21% in all players, 79.3% in Division 1, and 80.68% in Division 2. According to the result of decision trees, the top attribute of the Division 1 players was Theta wave, which was different from the result of other data in the Division 2 players. Also reveal the deference of brainwaves between the Division 2 players and the Division 1 players while they are in high stressed competitions.