Github user sscdotopen commented on the pull request:
https://github.com/apache/incubator-spark/pull/564#issuecomment-34583075
Centering a sparse matrix like it is done in textbook PCA is a serious
scalability bottleneck as it densifies the input matrix, not sure if you can
apply the
Github user sscdotopen commented on the pull request:
https://github.com/apache/incubator-spark/pull/575#issuecomment-34674441
I think making making the heavyweight mahout-core a dependency just for
access to the sparse vectors is no good idea. A better way would be to just
depend on
