I have unbalanced data - now what?

You can either perform cost-sensitive classification or resample your data to get a more balanced class distribution (see supervised Resample filter).