@Override
protected void setup(
final Reducer<Text, CountofDoubleWritable, GeoWaveOutputKey, DistortionEntry>.Context
context)
throws IOException, InterruptedException {
super.setup(context);
final ScopedJobConfiguration config =
new ScopedJobConfiguration(
context.getConfiguration(),
KMeansDistortionMapReduce.class,
KMeansDistortionMapReduce.LOGGER);
final int k = config.getInt(JumpParameters.Jump.COUNT_OF_CENTROIDS, -1);
if (k > 0) {
expectedK = k;
}
try {
centroidManager =
new CentroidManagerGeoWave<Object>(
context, KMeansDistortionMapReduce.class, KMeansDistortionMapReduce.LOGGER);
} catch (final Exception e) {
KMeansDistortionMapReduce.LOGGER.warn("Unable to initialize centroid manager", e);
throw new IOException("Unable to initialize centroid manager", e);
}
batchId =
config.getString(GlobalParameters.Global.PARENT_BATCH_ID, centroidManager.getBatchId());
}
@Override
public void reduce(
final Text key,
final Iterable<CountofDoubleWritable> values,
final Reducer<Text, CountofDoubleWritable, GeoWaveOutputKey, DistortionEntry>.Context
context)
throws IOException, InterruptedException {
double expectation = 0.0;
final List<AnalyticItemWrapper<Object>> centroids =
centroidManager.getCentroidsForGroup(key.toString());
// it is possible that the number of items in a group are smaller
// than the cluster
final Integer kCount;
if (expectedK == null) {
kCount = centroids.size();
} else {
kCount = expectedK;
}
if (centroids.size() == 0) {
return;
}
final double numDimesions = 2 + centroids.get(0).getExtraDimensions().length;
double ptCount = 0;
for (final CountofDoubleWritable value : values) {
expectation += value.getValue();
ptCount += value.getCount();
}
if (ptCount > 0) {
expectation /= ptCount;
final Double distortion = Math.pow(expectation / numDimesions, -(numDimesions / 2));
final DistortionEntry entry =
new DistortionEntry(key.toString(), batchId, kCount, distortion);
context.write(
new GeoWaveOutputKey(
DistortionDataAdapter.ADAPTER_ID, DistortionGroupManagement.DISTORTIONS_INDEX_LIST),
entry);
}
}