I've split my data and performed pre-processing. I ran some basic classifiers on it and got accuracies within 70-80%, which to me seems fairly low. One thing I didn't do was balance my classes before running all of these tests.
Is it alright if I run WEKA's class-balance filters on my training and data set after they are split, or do I have to do it before?
For reference, I have three sets at the moment, training, validation and test.