Odd Clustering Results
Hi all,
I receive some odd results when clustering.
My matrix file is meant to dipic two classes, a and b. Clustering (with default settings and 2 clusters) delivers the following solution:
cid Size ISim ISdev ESim ESdev Entpy Purty | a b
---------------------------------------------------------------
0 11 +1.000 +0.000 -1.000 +0.000 0.994 0.545 | 6 5
1 38 +1.000 +0.000 -1.000 +0.000 1.000 0.500 | 19 19
As I understand it the ISim = 1.0 in both clusters means that these objects are (as good as) identical, is that correct?
And if this is the case, then as a consequence the vectors within the matrix file do not represent the classes properly?
I'd appreciate your help!
Marie
Submitted by Marie on Wed, 2008-07-23 08:28
»
- Login to post comments
RE: Hey again, related to this I
Hey again, related to this I am wondering if Cluto can handle numbers in a format like:
-1.4829962E-4
Thanks again!
RE: Cluto should be able to read
Cluto should be able to read in numbers in scientific notation correctly. Can you use the flyspray issue tracking system to upload your files so that I can take a look.
RE: Hey, I am sorry it took me
Hey,
I am sorry it took me so long to answer. Unfortunatly my system broke down and I lost my data. I was not able to replicate the above result, nevertheless I am still receiving very bad clustering results:
2-way clustering: [I2=3.90e+001] [52 of 52], Entropy: 1.000, Purity: 0.500
------------------------------------------------------------------------
cid Size ISim ISdev ESim ESdev Entpy Purty | a b
------------------------------------------------------------------------
0 40 +0.612 +0.084 +0.396 +0.069 1.000 0.500 | 20 20
1 12 +0.414 +0.103 +0.396 +0.124 1.000 0.500 | 6 6
I'd still appreaciate any hint on what might go wrong.
Thank you again!!
Marie