Question

I’ve written a code to cluster a group of data in Java. I am using Apache Mahout to build clusters. Here is the piece of my code:

        Configuration conf = new Configuration();
        Path input = new Path("C:\\DATA\\input.txt");
        Path clusters = new Path("C:\\DATA\\clusters.txt");
        Path output = new Path("C:\\DATA\\output.txt");
        org.apache.mahout.common.distance.DistanceMeasure measure;
        String delta = new InterruptedException().toString();
        org.apache.mahout.clustering.kmeans.KMeansDriver myK = new KMeansDriver();
        myK.buildClusters(conf, input, clusters, output, measure, 100, delta, true);

But I get an error during creation of “measure” parameter : “variable measure might not have been initialized”.

Here is the doc of function buildClusters: https://builds.apache.org/job/Mahout-Quality/javadoc/org/apache/mahout/clustering/kmeans/KMeansDriver.html

So, how can I define the true value of “measure”?

Answer 1

DistanceMeasure (link) is just an Interface. You need to use one of the implementations of the interface, for example CosineDistanceMeasure or EuclideanDistanceMeasure

By the way, I wonder why your delta is defined like this: String delta = new InterruptedException().toString();

delta is defined as the convergence delta value, which internally is parsed as a double.