private void runMergeTrials()

in src/main/java/org/apache/datasketches/characterization/hll/DruidHllMergeAccuracyProfile.java [50:83]

31 lines of code
4 McCabe index (conditional complexity)


  private void runMergeTrials() {
    long key = random.nextLong();

    final int numTrials = Integer.parseInt(job.getProperties().mustGet("numTrials"));
    final int numSketches = Integer.parseInt(job.getProperties().mustGet("numSketches"));
    final int distinctKeysPerSketch = Integer.parseInt(job.getProperties().mustGet("distinctKeysPerSketch"));
    final double trueCount = numSketches * distinctKeysPerSketch;
    double sumEstimates = 0;
    double sumOfSquaredDeviationsFromTrueCount = 0;

    for (int t = 0; t < numTrials; t++) {
      final HyperLogLogCollector union = HyperLogLogCollector.makeLatestCollector();

      for (int s = 0; s < numSketches; s++) {
        final HyperLogLogCollector sketch = HyperLogLogCollector.makeLatestCollector();
        for (int k = 0; k < distinctKeysPerSketch; k++) {
          DruidHllAccuracyProfile.longToByteArray(key++, bytes);
          sketch.add(hash.hash(bytes));
        }
        union.fold(sketch);
      }
      final double estimatedCount = union.estimateCardinality();
      sumEstimates += estimatedCount;
      sumOfSquaredDeviationsFromTrueCount += (estimatedCount - trueCount) * (estimatedCount - trueCount);
    }
    final double meanEstimate = sumEstimates / numTrials;
    final double meanRelativeError = meanEstimate / trueCount - 1;
    final double relativeStandardError
      = Math.sqrt(sumOfSquaredDeviationsFromTrueCount / numTrials) / trueCount;
    job.println("True count: " + trueCount);
    job. println("Mean estimate: " + meanEstimate);
    job.println("Mean Relative Error: " + meanRelativeError);
    job.println("Relative Standard Error: " + relativeStandardError);
  }