gistfile1.scala

    // clusterization results
    val outputClustersPath = new Path("job/output-clusters")
    // textual dump of clusterization results
    val dumpPath = "job/dump"

    println("Running K-means...")
    // runs K-means algorithm with up to 20 iterations, to find clusters of colluding players (assumption of collusion is
    // made on the basis of number hand player together with other player[s])
    KMeansDriver.run(conf, vectorsPath, inputClustersPath, outputClustersPath,
      new CosineDistanceMeasure(), 0.01, 20, true, 0, false)

    println("Printing results...")

    // dumps clusters to a text file
    val clusterizationResult = finalClusterPath(conf, outputClustersPath, 20)
    val clusteredPoints = new Path(outputClustersPath, "clusteredPoints")
    val clusterDumper = new ClusterDumper(clusterizationResult, clusteredPoints)
    clusterDumper.setNumTopFeatures(10)
    clusterDumper.setOutputFile(dumpPath)
    clusterDumper.setTermDictionary(new Path(indexedDictionaryPath, "part-00000").toString,
      "sequencefile")
    clusterDumper.printClusters(null)