Statistical Analysis and Classification of EEG Signal

---

Abstract

Recently vast research on EEG signal is being explored in the area of machine learning. In this research, online EEGBCI data set is taken to analyze EEG signal. XDAWN denoising technique is implemented for signal pre-processing to improve signal to noise ratio (SNR) of EEG data. Then, filtering is done to separate different brain waves of EEG signal. After filtering, statistical tests are performed to extract most significant features present in the signal also, hypothesis tests have been performed to find statistically significant features. By performing the T-test, prominent features are extracted. Further, machine learning algorithms are used. Highest classification accuracy of 98.88% is obtained using KNN (K=3) classifier. The recent study is compared to the previous research on the same data set in which highest accuracy found was 97.77%. So, it can be concluded that, this research has improved methodology as compared to the reference research.

---

Keywords

ANOVA classification denoising EEG feature extraction filtering FIR