K-means clustering using Weka and WekaMine wrappers for Groovy

kmeansWM.groovy

#!/usr/bin/env groovy 

import durbin.weka.*
import static WM.*

dataFile = args[0]
k = args[1] as int

wm = new WM()
// Read a features x samples tab delimited file
data = wm.readNumericTab(dataFile)
// Many weka algorithms do not like string attributes, so remove string instance ID..
noIDinstances = wm.removeID(data)

// Create clusterer with a factory method...
km = SimpleKMeans(I:100,N:k,O:true)
km.buildClusterer(noIDinstances)

// Print out the original instance IDs along with cluster assignments
// data[instanceID][featureName] is some of the syntax sugar that comes
// with WekaMine
noIDinstances.eachWithIndex{instance,i->
	println "${data[i]['ID']}\t"+km.clusterInstance(instance)
}