oops. I figured it out. Please specify a -k (number of cluster) parameter and the distance threshold. :) KMeans need to know either the cluster count or the clusters in the -c clusters folder. If it doesn't find k then it assumes you have initial clusters put in the clusters folder.
PS: you can simply do a "bin/mahout kmeans" to run kmeans Since Mahout 0.3 Robin On Sat, May 8, 2010 at 2:47 PM, david.stu...@progressivealliance.co.uk <david.stu...@progressivealliance.co.uk> wrote: > Hi Robin, > > I'm using the latest from trunk so 0.4 > > Seq file dump > > bin/mahout org.apache.mahout.utils.SequenceFileDumper --seqFile /tmp/out.txt > Input Path: /tmp/out.txt > Key class: class org.apache.hadoop.io.LongWritable Value Class: class > org.apache.mahout.math.VectorWritable > Key: 0: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > Key: 1: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > Key: 2: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > Key: 3: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > Key: 4: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > Key: 5: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > Key: 6: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > Key: 7: Value: org.apache.mahout.math.vectorwrita...@6bffc686 > > > > On 08 May 2010 at 10:57 Robin Anil <robin.a...@gmail.com> wrote: > >> David, couple of things needed to debug this >> 1) Tell me which version of Mahout are you using. >> 2) use o.a.m.utils.SequenceFileDumper to dump the out.txt and see what >> the key and value classes are >> >> Robin
