log.info("Loaded: " + clusters.size() + " clusters");
}
private static void runMinHash(Configuration conf, Path samples, Path output)
throws Exception {
MinHashDriver mhd = new MinHashDriver();
ToolRunner.run(conf, mhd, new String[] { "--input", samples.toString(),
"--hashType", HashFactory.HashType.MURMUR3.toString(), "--output",
output.toString(), "--minVectorSize", "1", "--debugOutput"