On Wed, Oct 24, 2012 at 2:32 AM, Dan Filimon <[email protected]>wrote:
> So, the way I'm thinking of doing it is: > 1. refactor (and prettify?) the existing code to support adding new > points directly (both weighted and unweighted) and update the tests; > 2. write the wrapper CentroidWritable class and the MapReduce itself > using the new API from (1). > Yes. These sound good. > I also know that our goal is to integrate the code into Mahout so > maybe it might make some sense to pass around Clusters rather than > Centroids? While this makes sense, I feel that for now, not hooking > into the more complex clustering API for this first prototype would be > more straightforward. > This also sounds very good. Working and simple first. Fully integrated and compatible after. (sorry to be terse, I am time crunched today)
