Analysis on Non-Linear Features of Electroencephalogram (EEG) Signal for Neuromarketing Application

Neuromarketing is thestudy of the human brain's response towards commercial, brands and other marketing stimuli. The experimental protocol consists of four categories of merchandises with five products each. The eego Sports device (ANT Neuro, Enschede, The Netherlands) with 32 channels at a sampling frequency of 512 Hz are used for data collection from 10 subjects. The raw EEG signal is bandpass filtered into two frequency bands which are alpha (8-13 Hz) and beta (13-30 Hz). The non-linear feature, Detrended Fluctuation Analysis (DFA) are extracted from filtered signals. Twoclassifiers, Neural Network (NN) and k-Nearest Neighbors (k-NN) are used to analyse the features extracted. 1s segments of EEG signals are used in this study. DFA features indicate that the Fast Food category is most preferred as it obtained the highest classification accuracy of 80%. The most preferred products for Category A (Smartphones) is iPhone, Category B (Automobiles) is Honda, Category C (Fast Food) is McDonalds and Category D is (Beverages) is milk. The DFA value for alpha waves are highest and for beta waves are lowest if the subjects prefer the products.