Github user fmcquillan99 commented on the issue: https://github.com/apache/madlib/pull/315 Actually the earlier issue above ^^^ is OK, where I said `I'm not sure what this is doing` because forcing all training data to be a single point means that the distance to all test points is identical. So a random nearest neighbor could be picked. Which it is what seems to be happening. So I think just fix the error above and this should be good to go.
---