weka - Trained and Test data have different number of attributes that gave an error "Train and test set are not compatible"

Question

I use WEKA for Text classification , I have trained data set , and I apply StringToWOrdVector and NumericToNominal filters , and have test data set and applied the same filters on it . When I try to apply my model on test data ,it gave me the following error Train and test set are not compatible I searched for a solution , the error occurred because number of attributes different between two sets, and it always be different because texts in two sets are different

How I can solve this error please ?

Answer 1

The best thing you can do is combine your training and test set into one file and then apply the filter to it all in one go, then split them up again and copy the @attribute values from the combined file into both the training and test files. This way the attributes will be consistent across both files.