Given the estimated frequencies of a join key in two pipes that we want to skew-join together, this returns the key's replication amount in each pipe.
Given the estimated frequencies of a join key in two pipes that we want to skew-join together, this returns the key's replication amount in each pipe.
Note: if we switch to a Count-Min sketch, we'll need to change the meaning of these counts from "sampled counts" to "estimates of full counts", and also change how we deal with counts of zero.
See https://github.com/twitter/scalding/pull/229#issuecomment-10792296