The current classifier pipeline takes a long time to fit and may be fitting the same model multiple times. This should be looked at in hopes of finding some low-hanging fruit in performance gains.