how to use cluto to cluster high-dimension data
Dear Prof. Karypis,
Thanks for your excellent tool firstly.
Now, I try to use cluto-2.1.1 to help me do clustering high-dimension data.
I have some questions to consult you:
1. one of my trial is:
"vcluster.exe -sim=jacc -clmethod=graph myData_wgn.txt 5"
But the clustering result has six (not five) clustering index: "-1,0,1,2,3,4", i do not know the "-1" means what.
2. my data is about 39 dims to 256 dims, i wonder to know whether they are real high-dimension data or not?
3. is there some questions if i use Kmeans to cluser my data (about 39 dims to 256dims)?
4. if i want to use cluto to cluster my data (about 39 dims to 256dims), what is the best parameters combination? CLMethod=GRAPH ??? SimFun=dist ???
thanks a lot.
Submitted by xuyong on Thu, 2013-01-10 05:21
»
- Login to post comments