Apply initial centroids on Spark Kmeans workload.#187

Open

pfxuan wants to merge 1 commit intoIntel-bigdata:masterfrom pfxuan:spark-kmeans

+34-25

Commits on Mar 30, 2016

Apply initial centroids on Spark Kmeans workload, which makes HiBench hold the same convergence condition on both MapReduce and Spark benchmark.
pfxuan
committed