HTM prediction model with good accuracy but low activation levels

[jclevesque]

Thanks for the pointers. I was able to get the active columns from the spatial pooler by taking:

sp = model._getSPRegion.getSelf()
activeColumns = sp._spatialPoolerOutput