I have tried converting all the files (training, labeled test data and un-labeled test data) prefereably in arff format and it is working good for my labelled test data. But for unlabelled test data it keep poping the error (files are not compatible or different number of labels).
Here is what I tried: I added "?" for missing values in the "class label" column of un-labelled test data and also converted the type of "class label" attribute from string to nominal using weka filters (because intitially the error was "str not compatible to nom"). However, it is still not working to make predictions of un-labelled test data and still giving the error (different nummber of labels). In my opinion I am doing it in a wrong way, Can someone tell me where am I wrong or any suggestion to do it in correct way? Thank you very much