A Comprehensive Performance Study of Classification of Emotions using EEG Centroid Data

Muhammad Umair, Syed Sajjad, Hussain Rizvi, Kamran Raza
Classification of human emotion using EEG signal is an interesting research topic for many. In this paper a detailed study of classification efficiency of different machine learning algorithms using EEG signal was done. The study make use of DEAP data set. The Valance, Arousal and Dominance values reported in DEAP data set were ranked on Mehrabian 3D emotion classification model to produce 8 emotions. k-mean was used to reduce number of observations in original data set and 3, 10 and 40
more » ... , 10 and 40 centroids were calculated for EEG data. 11 different combinations of emotions were tested using 20 different variations of Decision Trees, SVM, kNN and Ensemble classifiers. This study is one of its kind which uses k-mean as number of observation reducing technique and rank top performing algorithm for each set of emotion..