[dododo]

there's nothing wrong with looking at some of the data first per se, provided you do not use the same data to draw conclusions. i.e., have a training and a test data set.