EEG Signal Analysis Using PCA and Logistic Regression

Abstract

The analysis of brain signals, known as EEG, is a field of great interest, especially in rehabilitation medicine. For this area, the development of Brain Computer Interface (BCI) systems is of great value since it offers a new channel of communication between the individual and external systems through the analysis of brain signals. This work aims to analyze brain signals used in Brain Computer Interface systems using one of the most widespread techniques for such, Principal Component Analysis (PCA). An online database (Graz data set B) containing EEG signals of motor imagery activity of left and right hand was used. These signals were analyzed and their characteristics were extracted using PCA. To identify them according to the motor imagery performed, these characteristics were classified using a classification algorithm, Logistic Regression.