private Instance createInstance(String id, Double value) {
Instance inst = new Instance(id);
NumericVectorFeature feature = new NumericVectorFeature("num");
feature.setValue(new double[] {value});
inst.addFeature(feature);
return inst;
}
/** @param args */
public static void main(String[] args) {
DataSet ds = new DataSet();
Instance inst = new Instance("1");
StringFeature feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jack black");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("san diego");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("2");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jack black");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("3");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jack");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("san diego");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("4");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jack l. black");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("san diego");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("5");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("j. black");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("san diego");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("6");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("j black");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("new york");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("7");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("black");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("new york");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("8");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jackie black");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("9");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jack brown");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("10");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jackie green");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("11");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("bob");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("san fran");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("12");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("bobbie");
inst.addFeature(feature);
feature = new StringFeature(FEATURE_NAME2);
feature.setValue("san fran");
inst.addFeature(feature);
ds.add(inst);
inst = new Instance("13");
feature = new StringFeature(FEATURE_NAME1);
feature.setValue("jackie");
inst.addFeature(feature);
ds.add(inst);
DPMeans clusterer = new DPMeans(5, true);
clusterer.registerFeatureType(FEATURE_NAME1, StringMedianCentroid.class, new EditDistance(0.5));
clusterer.registerFeatureType(
FEATURE_NAME2, StringMedianCentroid.class, new ExactTokenMatchDistance(0.5));
clusterer.setThreshold(0.5);
ClusterResult clusters = clusterer.doCluster(ds);
for (Cluster c : clusters) {
System.out.println(c.toString(true));
}
clusterer.terminate();
}