I don't see them, no, unless I'm misunderstanding something. We have cosine, mahalanobis, tanimoto, Euclidean, Manhattan, and some variants like squared and weighted. I think these additions are fine?
On Fri, Dec 31, 2010 at 11:51 PM, Ted Dunning <[email protected]> wrote: > Don't we have both of these already under different names? > > > On Friday, December 31, 2010, Lance Norskog (JIRA) <[email protected]> wrote: >> Minkowski and Chebyshev DistanceMeasure >> --------------------------------------- >> >> Key: MAHOUT-571 >> URL: https://issues.apache.org/jira/browse/MAHOUT-571 >> Project: Mahout >> Issue Type: New Feature >> Components: Math >> Reporter: Lance Norskog >> Priority: Minor >> Attachments: MAHOUT-571.patch >> >> Implementations of Minkowski and Chebyshev distance measures. >> Minkowski distance is a generalization of the L-space measures, where L1 is >> Manhattan distance and L2 is Euclidean distance. Uses Math.pow to calculate >> coordinate distances. Math.pow has a fast-path for integer-valued arguments, >> so Minkowski with 3.0 is much faster than Minkowski with 3.1. >> >> Chebyshev distance is "chessboard" distance, based on the moves that a king >> can make: any direction include diagonals. The Manhattan or taxicab >> distances can only traverse in right angles. >> >> -- >> This message is automatically generated by JIRA. >> - >> You can reply to this email to add a comment to the issue online. >> >> >
