Then, all the information was integrated in a single dataset
and it was saved in the .ARFF format of Weka [14]. Next, the
whole dataset was divided randomly into 10 pairs of training
and test data files (maintaining the original class distribution).
This way, each classification algorithm can be evaluated using
stratified tenfold cross-validation. So after preprocessing we
have a dataset with 77 attributes/variables of 670 students.