I agree, they're not efficient to apply and aren't frequently applied these days. You'd more usually pick techniques that aren't sensitive to popularity effects and mean/stdev of data.
On Thu, Jun 13, 2013 at 8:53 PM, Sebastian Schelter <[email protected]> wrote: > I propose to also deprecate the similarity and preference transforms > from org.apache.mahout.cf.taste.transforms and > org.apache.mahout.cf.taste.impl.transforms as I don't recall anyone ever > using them. > > Any objections? > > -sebastian
