Suppose we have dataset of 9 subjects of binary motor imagery classification and we want to test our machine learning algorithm on that. Then how should we proceed ?
Should we train and test on the subjects data separately or should we combine the data from the subjects and then train/test them? How to proceed?
Hello,
it really depends on your use case and research question … but a commonly applied approach is to train & test the classifier for each participant separately. To generate the “overall” performance, one calculates the average classification score across participants.