Fixes multilabel classifiers and multilabel generator (#249)

* Fixes multilabel classifiers and multilabel generator

* Formatting

moa/src/main/java/com/yahoo/labs/samoa/instances/Range.java

@@ -27,7 +27,7 @@ public class Range implements Serializable {
 
     public Range(String range) {
         this.rangeText = range;
-        //this.setRange(range); //needs upperLimit
+        this.setRange(range); //needs upperLimit
     }
 
     /**

moa/src/main/java/moa/classifiers/multilabel/MEKAClassifier.java

@@ -48,7 +48,7 @@
  * @author Jesse Read
  * @version $Revision: 1 $
  */
-public class MEKAClassifier extends AbstractMultiLabelLearner implements MultiTargetRegressor, Serializable {
+public class MEKAClassifier extends AbstractMultiLabelLearner implements MultiLabelLearner, MultiTargetRegressor, Serializable {
 
 	private static final long serialVersionUID = 1L;
 

moa/src/main/java/moa/classifiers/multilabel/MajorityLabelset.java

@@ -28,6 +28,7 @@
 import com.yahoo.labs.samoa.instances.MultiLabelPrediction;
 import com.yahoo.labs.samoa.instances.Prediction;
 import moa.classifiers.AbstractMultiLabelLearner;
+import moa.classifiers.MultiLabelLearner;
 import moa.classifiers.MultiTargetRegressor;
 import moa.core.StringUtils;
 
@@ -38,7 +39,7 @@
  * @author Jesse Read ([email protected])
  * @version $Revision: 1 $
  */
-public class MajorityLabelset extends AbstractMultiLabelLearner implements MultiTargetRegressor {
+public class MajorityLabelset extends AbstractMultiLabelLearner implements MultiLabelLearner {
     //AbstractClassifier {
 
     private static final long serialVersionUID = 1L;

moa/src/main/java/moa/classifiers/multilabel/MultilabelHoeffdingTree.java

@@ -98,11 +98,13 @@ public MultilabelLearningNodeClassifier(double[] initialClassObservations, Class
 			super(initialClassObservations);
 
 			if (cl== null) {
-				this.classifier = ((Classifier) getPreparedClassOption(ht.learnerOption)).copy();
-				this.classifier.resetLearning();
+				MEKAClassifier learner = new MEKAClassifier();
+				learner.baseLearnerOption.setValueViaCLIString("meka.classifiers.multilabel.incremental.PSUpdateable");
+				learner.prepareForUse();
+				learner.setModelContext(ht.getModelContext());
 
-				InstancesHeader raw_header = ht.getModelContext();
-				this.classifier.setModelContext(raw_header);
+				this.classifier = learner;
+				this.classifier.resetLearning();
 			}
 			else{
 				this.classifier = cl.copy();

moa/src/main/java/moa/classifiers/multilabel/trees/ISOUPTree.java

@@ -1481,4 +1481,4 @@ public void log(String s) {
 	// endregion --- Processing methods
 
 	// endregion ================ METHODS ================
-}
+}

moa/src/main/java/moa/streams/generators/multilabel/MetaMultilabelGenerator.java

@@ -28,11 +28,14 @@
 import com.github.javacliparser.FloatOption;
 import com.github.javacliparser.IntOption;
 import moa.streams.InstanceStream;
+import moa.streams.MultiTargetInstanceStream;
 import moa.tasks.TaskMonitor;
 import com.yahoo.labs.samoa.instances.Attribute;
+import com.yahoo.labs.samoa.instances.DenseInstance;
 import com.yahoo.labs.samoa.instances.Instance;
 import com.yahoo.labs.samoa.instances.Instances;
 import com.yahoo.labs.samoa.instances.InstancesHeader;
+import com.yahoo.labs.samoa.instances.Range;
 import com.yahoo.labs.samoa.instances.SparseInstance;
 import moa.core.FastVector;
 import moa.core.Utils;
@@ -43,7 +46,7 @@
  * @author Jesse Read (([email protected]))
  * @version $Revision: 7 $
  */
-public class MetaMultilabelGenerator extends AbstractOptionHandler implements InstanceStream {
+public class MetaMultilabelGenerator extends AbstractOptionHandler implements MultiTargetInstanceStream {
 
     private static final long serialVersionUID = 1L;
 
@@ -153,19 +156,25 @@ public void restart() {
      * @param	si	single-label Instances
      */
     protected MultilabelInstancesHeader generateMultilabelHeader(Instances si) {
-        Instances mi = new Instances(si, 0, 0);
-        mi.setClassIndex(-1);
-        mi.deleteAttributeAt(mi.numAttributes() - 1);
-        FastVector bfv = new FastVector();
-        bfv.addElement("0");
-        bfv.addElement("1");
-        for (int i = 0; i < this.m_L; i++) {
-            mi.insertAttributeAt(new Attribute("class" + i, bfv), i);
-        }
-        this.multilabelStreamTemplate = mi;
-        this.multilabelStreamTemplate.setRelationName("SYN_Z" + this.labelCardinalityOption.getValue() + "L" + this.m_L + "X" + m_A + "S" + metaRandomSeedOption.getValue() + ": -C " + this.m_L);
-        this.multilabelStreamTemplate.setClassIndex(this.m_L);
-        return new MultilabelInstancesHeader(multilabelStreamTemplate, m_L);
+    	Instances mi = new Instances(si, 0, 0);
+    	mi.deleteAttributeAt(mi.numAttributes() - 1);
+    	FastVector bfv = new FastVector();
+    	bfv.addElement("0");
+    	bfv.addElement("1");
+    	for (int i = 0; i < this.m_L; i++) {
+    		mi.insertAttributeAt(new Attribute("class" + i, bfv), i);
+    	}
+
+    	Range range = new Range(Integer.toString((numLabelsOption.getValue())));
+
+    	this.multilabelStreamTemplate = mi;
+    	this.multilabelStreamTemplate.setRelationName("SYN_Z" + this.labelCardinalityOption.getValue() + "L" + this.m_L + "X" + m_A + "S" + metaRandomSeedOption.getValue() + ": -C " + this.m_L);
+    	this.multilabelStreamTemplate.setClassIndex(Integer.MAX_VALUE);
+    	this.multilabelStreamTemplate.setRangeOutputIndices(range);
+
+    	MultilabelInstancesHeader header = new MultilabelInstancesHeader(multilabelStreamTemplate, m_L);
+    	header.setRangeOutputIndices(range);
+    	return header;
     }
 
     /**
@@ -267,7 +276,7 @@ private double joint(int k, int y[]) {
     private Instance generateMLInstance(HashSet<Integer> Y) {
 
         // create a multi-label instance:
-        Instance x_ml = new SparseInstance(this.multilabelStreamTemplate.numAttributes());
+        Instance x_ml = new DenseInstance(this.multilabelStreamTemplate.numAttributes());
         x_ml.setDataset(this.multilabelStreamTemplate);
 
         // set classes
@@ -472,7 +481,7 @@ public int compare(HashSet Y1, HashSet Y2) {
         }
 
         // shuffle 
-        Collections.shuffle(Arrays.asList(map_set));
+        Collections.shuffle(Arrays.asList(map_set), m_MetaRandom);
 
         // return
         return map_set;
@@ -545,7 +554,7 @@ private ArrayList<Integer> getShuffledListToLWithoutK(int L, int k) {
                 list.add(j);
             }
         }
-        Collections.shuffle(list);
+        Collections.shuffle(list, m_MetaRandom);
         return list;
     }