use of de.lmu.ifi.dbs.elki.result.EvaluationResult in project elki by elki-project.
the class EvaluateSquaredErrors method evaluateClustering.
/**
* Evaluate a single clustering.
*
* @param db Database
* @param rel Data relation
* @param c Clustering
* @return ssq
*/
public double evaluateClustering(Database db, Relation<? extends NumberVector> rel, Clustering<?> c) {
boolean square = !distance.isSquared();
int ignorednoise = 0;
List<? extends Cluster<?>> clusters = c.getAllClusters();
double ssq = 0, sum = 0;
for (Cluster<?> cluster : clusters) {
if (cluster.size() <= 1 || cluster.isNoise()) {
switch(noiseOption) {
case IGNORE_NOISE:
ignorednoise += cluster.size();
continue;
case TREAT_NOISE_AS_SINGLETONS:
continue;
case MERGE_NOISE:
// Treat as cluster below:
break;
}
}
NumberVector center = ModelUtil.getPrototypeOrCentroid(cluster.getModel(), rel, cluster.getIDs());
for (DBIDIter it1 = cluster.getIDs().iter(); it1.valid(); it1.advance()) {
final double d = distance.distance(center, rel.get(it1));
sum += d;
ssq += square ? d * d : d;
}
}
final int div = Math.max(1, rel.size() - ignorednoise);
if (LOG.isStatistics()) {
LOG.statistics(new DoubleStatistic(key + ".mean", sum / div));
LOG.statistics(new DoubleStatistic(key + ".ssq", ssq));
LOG.statistics(new DoubleStatistic(key + ".rmsd", FastMath.sqrt(ssq / div)));
}
EvaluationResult ev = EvaluationResult.findOrCreate(db.getHierarchy(), c, "Internal Clustering Evaluation", "internal evaluation");
MeasurementGroup g = ev.findOrCreateGroup("Distance-based Evaluation");
g.addMeasure("Mean distance", sum / div, 0., Double.POSITIVE_INFINITY, true);
g.addMeasure("Sum of Squares", ssq, 0., Double.POSITIVE_INFINITY, true);
g.addMeasure("RMSD", FastMath.sqrt(ssq / div), 0., Double.POSITIVE_INFINITY, true);
db.getHierarchy().add(c, ev);
return ssq;
}
Aggregations