forum.alglib.net
http://forum.alglib.net/

K-means clustering: finding out number of clustered elements
http://forum.alglib.net/viewtopic.php?f=2&t=2248
Page 1 of 1

Author:  maksims [ Sun Feb 22, 2015 7:34 pm ]
Post subject:  K-means clustering: finding out number of clustered elements

I'm trying to create a "heat map" from a large dataset of coordinates, where clustering fits just fine. The k-means++ clustering example works OK, but I need not only mean coordinates based on K (stored in kmeansreport.c array), but also total number of original elements for each resulting coordinate:

Code:
Dim xy(4, 1) As Double
xy(0, 0) = 1 : xy(0, 1) = 1
xy(1, 0) = 1 : xy(1, 1) = 2
xy(2, 0) = 4 : xy(2, 1) = 1
xy(3, 0) = 2 : xy(3, 1) = 3
xy(4, 0) = 4 : xy(4, 1) = 1.5

Dim s As alglib.clusterizerstate = Nothing
Dim rep As alglib.kmeansreport = Nothing
alglib.clusterizercreate(s)
alglib.clusterizersetpoints(s, xy, 2)
alglib.clusterizersetkmeanslimits(s, 5, 0)
alglib.clusterizerrunkmeans(s, 2, rep)

Now it's possible to inspect rep.c and find out resulting coordinate pairs:

Code:
System.Console.WriteLine("X=" + rep.c(0, 0).ToString + ", Y=" + rep.c(0, 1).ToString)
System.Console.WriteLine("X=" + rep.c(1, 0).ToString + ", Y=" + rep.c(1, 1).ToString)

But, how do I compute total number of original xy elements "behind" each rep.c coordinate? When I create a heat map I need x,y and "weight" value - how do I find one?

Regards,
Maksim Sestic

Page 1 of 1 All times are UTC
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/