I want to predict and distribute brain waves with a naive Bayes classifier.

I made a matrix of 7500 vertical and 1000 horizontal by applying the data brought by openBCI to the Han window and fft. This matrix is an electroencephalogram when you act differently by 2500 in the vertical direction. I want to predict and distribute this with a naive Bayes classifier. So I have a question. First, how should I label it? How should the predicted distribution be made after the second labeling?