public static void main(String[] args) {
if (args.length < 2) {
System.err.println("Usage: KMeansMP <input_file> <results>");
System.exit(1);
}
String inputFile = args[0];
String results_path = args[1];
JavaPairRDD<Integer, Iterable<String>> results;
int k = 4;
int iterations = 100;
int runs = 1;
long seed = 0;
final KMeansModel model;
SparkConf sparkConf = new SparkConf().setAppName("KMeans MP");
JavaSparkContext sc = new JavaSparkContext(sparkConf);
JavaRDD<String> lines = sc.textFile(inputFile);
JavaRDD<Vector> points = lines.map(new ParsePoint());
JavaRDD<String> titles = lines.map(new ParseTitle());
model = KMeans.train(points.rdd(), k, iterations, runs, KMeans.RANDOM(), 0);
results = titles.zip(points).mapToPair(new ClusterCars(model)).groupByKey();
results.saveAsTextFile(results_path);
sc.stop();
}
/**
* This method trains a k-means clustering model
*
* @param data Training data as a JavaRDD of Vectors
* @param noOfClusters Number of clusters
* @param noOfIterations Number of iterations to run
* @param noOfRuns Number of runs of the algorithm to execute in parallel
* @param initializationMode Initialization algorithm: random or k-means||
* @return
*/
public KMeansModel train(
JavaRDD<Vector> data,
int noOfClusters,
int noOfIterations,
int noOfRuns,
String initializationMode) {
return org.apache.spark.mllib.clustering.KMeans.train(
data.rdd(), noOfClusters, noOfIterations, noOfRuns, initializationMode);
}
