A Mobile Application for Estimating Emotional Valence Using a Single-Channel EEG Device

A product assessment is the important process to develop a new product. After a new product has been developed, the product developers hire ordinary people and give an interview to them. In recent years, a new method called 'neuromarketing' is used for product evaluation. However, it is difficult to use the conventional measurement devices and they are mainly used in an experimental environment. In this paper, we developed the model to estimate human emotions, especially valence by using single-channel EEG device. We used the fast Fourier transform, the robust scaling and the support vector regression to predict the valence score. The parameters of the methods were selected by using the grid search and the genetic algorithm. The designed model was evaluated by the correlation coefficient and the classification accuracy of two classes between predicted valence data and labeled valence data. The scores were 0.36 and 72.40%.