Different results while running a EM clustering process on RapidMiner and weka

lokraj

I am running the EM clustering for the same data using rapidminer and weka for the same parameters. But I found a different result. Here is my process,  data, and result. Why this happens does anyone explain? Thanks in advance.
rapidminer result

Cluster 0: 1108 items
Cluster 1: 0 items
Cluster 2: 0 items
Total number of items: 1108

weka result

Clustered Instances

0       114 ( 10%)

1       813 ( 73%)

2       181 ( 16%)

sgenzer

hi @lokraj so getting different results does not surprise me. k-means clustering uses a local random seed to start the process. I don't see your Weka process so I cannot compare but possibly you're seeing randomness at play.

Scott