If your aim is performing classification/regression techniques by building a prediction model from training data and then apply this model on the test data, then yes, you can do that. But the design structure of both data should be compatible in terms of attributes' name, order and type. 

Please let us know if you still have further issues.

Regards, 
Martin


View this message in context: Re: two data sets
Sent from the WEKA mailing list archive at Nabble.com.