Wavelet based-analysis of alpha rhythm on EEG signal

One of the major frequency rhythm in EEG signal is called alpha rhythm, that indicate relax condition, calm, and awake without much concentration. In this paper we analyzing alpha rhythm using continuous wavelet transform (CWT) to explore the feature of relax condition. We do some scenario in analyzing alpha rhythm, normalizing and segmenting the data. EEG dataset was provided by DEAP. We sort the relax data (labelled with high valence and low arousal by participants) among all data to be observed. First, EEG data are normalized then filtered using band pass filter to get the specific alpha frequency (8–13Hz). Then, we use CWT to transform the signals into time-frequency domain. Entropy and energy of the coefficient wavelet transform are calculate as feature for clustering. From the result, normalized data gave different values. Besides changes the real magnitude information, it give lower accuracy 51.7% than not normalized data 67.2%. We conclude that normalizing data is not necessary especially on subject independent analysis. In additional, clustering result of all data compared with segmented data aren't gave significant differences. Finally, using CWT for feature extraction gives good enough results (67.2%).