Decision tree method to classify the electroencephalography-based emotion data

Abstract

Electroencephalography (EEG) data contains recordings of brain signal activity divided into several channels with different impulse responses that can be used to detect human emotions. In classifying emotions, EEG data needs to be parsed or signal processed into values ​​that can help recognize emotions. Research related to electroencephalography has been carried out previously and has experienced success using the Fuzzy C-Means, Multiple Discriminant Analysis, and Deep Neural Network methods. This study was conducted to classify human emotions from electroencephalography data on 10 participants. Each participant carried out 40 trials of testing using the Power Spectral Density (PSD) and Discrete Wavelet Transform (DWT) methods at the initial stage of classification and the Decision Tree method as the final method that can improve the accuracy of the two methods at the initial stage of classification. The results of this study were the finding of 2 participants (3 trials) who were unmatched from a total of 10 participants (400 trials), which were analyzed using the decision tree method. The decision tree method can correct this error and increase the classification result to 100%. The DWT method is used as a reference in the classification of emotions, considering that the DWT method has an output of arousal and valance values ​​. In contrast, the PSD method only has a combined output.