How is the output of Fisher Linear Discriminant Analysis experiment interpreted now that the column labels in the output are replaced with Col1, Col2, Col3.......etc? How can the model be used to predict clusters of other input data as deployed web service requires even the dependent valuable(the same same ones we wish to predict)?