remove generics from model interfaces and results to make manipulatio…

…n much clearer, generics are now hidden inside abstract base classes

README.md

@@ -50,7 +50,6 @@ notebooks or to document the idea you are working on. You have to install jupyte
 on that you can follow the instruction from [here](https://github.com/SpencerPark/IJava#installing). The following notation is
  specific to IJava kernel jupyter notation. 
 
-    %mavenRepo oss-sonatype-snapshots https://oss.sonatype.org/content/repositories/snapshots/
     %maven io.github.padreati:rapaio:2.2.2    
 
 The last option to use the library is do download the release files from this repository. If you use IntelliJ Idea IDE, you can use 

src/rapaio/experiment/ml/classifier/boost/AdaBoostSAMME.java

@@ -51,9 +51,7 @@
  * <p>
  * User: Aurelian Tutuianu <[email protected]>
  */
-public class AdaBoostSAMME
-        extends AbstractClassifierModel<AdaBoostSAMME, ClassifierResult<AdaBoostSAMME>>
-        implements Printable {
+public class AdaBoostSAMME extends AbstractClassifierModel<AdaBoostSAMME, ClassifierResult> implements Printable {
 
     private static final long serialVersionUID = -9154973036108114765L;
     private static final double delta_error = 10e-10;
@@ -199,8 +197,8 @@ private boolean learnRound(Frame df) {
     }
 
     @Override
-    protected ClassifierResult<AdaBoostSAMME> corePredict(Frame df, boolean withClasses, boolean withDistributions) {
-        ClassifierResult<AdaBoostSAMME> fit = ClassifierResult.build(this, df, withClasses, true);
+    protected ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDistributions) {
+        ClassifierResult fit = ClassifierResult.build(this, df, withClasses, true);
         for (int i = 0; i < h.size(); i++) {
             ClassifierResult hp = h.get(i).predict(df, true, false);
             for (int j = 0; j < df.rowCount(); j++) {

src/rapaio/experiment/ml/classifier/boost/GBTClassifierModel.java

@@ -52,7 +52,7 @@
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> at 12/12/14.
  */
 public class GBTClassifierModel
-        extends AbstractClassifierModel<GBTClassifierModel, ClassifierResult<GBTClassifierModel>>
+        extends AbstractClassifierModel<GBTClassifierModel, ClassifierResult>
         implements Printable {
 
     private static final long serialVersionUID = -2979235364091072967L;
@@ -191,22 +191,22 @@ private void buildAdditionalTree(Frame df, Var w, DMatrix yk) {
             tree.fit(train, sample.weights, "##tt##");
             trees.get(k).add(tree);
 
-            RegressionResult<RTree> rr = tree.predict(df, false);
+            RegressionResult rr = tree.predict(df, false);
             for (int i = 0; i < df.rowCount(); i++) {
                 f.set(k, i, f.get(k, i) + shrinkage * rr.firstPrediction().getDouble(i));
             }
         }
     }
 
     @Override
-    public ClassifierResult<GBTClassifierModel> corePredict(Frame df, boolean withClasses, boolean withDistributions) {
-        ClassifierResult<GBTClassifierModel> cr = ClassifierResult.build(this, df, withClasses, withDistributions);
+    public ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDistributions) {
+        ClassifierResult cr = ClassifierResult.build(this, df, withClasses, withDistributions);
 
         DMatrix p_f = SolidDMatrix.empty(K, df.rowCount());
 
         for (int k = 0; k < K; k++) {
             for (RTree tree : trees.get(k)) {
-                RegressionResult<RTree> rr = tree.predict(df, false);
+                RegressionResult rr = tree.predict(df, false);
                 for (int i = 0; i < df.rowCount(); i++) {
                     p_f.set(k, i, p_f.get(k, i) + shrinkage * rr.firstPrediction().getDouble(i));
                 }

src/rapaio/experiment/ml/classifier/ensemble/CForest.java

@@ -54,7 +54,6 @@
 import rapaio.ml.common.Capabilities;
 import rapaio.ml.common.VarSelector;
 import rapaio.ml.eval.metric.Confusion;
-import rapaio.printer.Printable;
 import rapaio.printer.Printer;
 import rapaio.printer.opt.POption;
 import rapaio.util.Pair;
@@ -75,9 +74,7 @@
  * <p>
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> on 4/16/15.
  */
-public class CForest
-        extends AbstractClassifierModel<CForest, ClassifierResult<CForest>>
-        implements Printable {
+public class CForest extends AbstractClassifierModel<CForest, ClassifierResult> {
 
     private static final long serialVersionUID = -145958939373105497L;
 
@@ -462,12 +459,12 @@ private Pair<ClassifierModel, VarInt> buildWeakPredictor(Frame df, Var weights)
     }
 
     @Override
-    protected ClassifierResult<CForest> corePredict(Frame df, boolean withClasses, boolean withDensities) {
-        ClassifierResult<CForest> cp = ClassifierResult.build(this, df, true, true);
-        var treeFits = predictors.stream().parallel()
-                .map(pred -> pred.predict(df, baggingMode.needsClass(), baggingMode.needsDensity()))
+    protected ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDensities) {
+        ClassifierResult cp = ClassifierResult.build(this, df, true, true);
+        List<ClassifierResult> treeFits = predictors.stream().parallel()
+                .map(pred -> (ClassifierResult) pred.predict(df, baggingMode.needsClass(), baggingMode.needsDensity()))
                 .collect(Collectors.toList());
-        baggingMode.computeDensity(firstTargetLevels(), new ArrayList<>(treeFits), cp.firstClasses(), cp.firstDensity());
+        baggingMode.computeDensity(firstTargetLevels(), treeFits, cp.firstClasses(), cp.firstDensity());
         return cp;
     }
 

src/rapaio/experiment/ml/classifier/meta/CBinaryLogisticStacking.java

@@ -53,7 +53,7 @@
  * <p>
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> on 9/30/15.
  */
-public class CBinaryLogisticStacking extends AbstractClassifierModel<CBinaryLogisticStacking, ClassifierResult<CBinaryLogisticStacking>> implements Printable {
+public class CBinaryLogisticStacking extends AbstractClassifierModel<CBinaryLogisticStacking, ClassifierResult> implements Printable {
 
     private static final long serialVersionUID = -9087871586729573030L;
 
@@ -164,7 +164,7 @@ protected PredSetup preparePredict(Frame df, boolean withClasses, boolean withDi
     }
 
     @Override
-    protected ClassifierResult<CBinaryLogisticStacking> corePredict(Frame df, boolean withClasses, boolean withDistributions) {
+    protected ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDistributions) {
         return ClassifierResult.copy(this, df, withClasses, withDistributions, log.predict(df));
     }
 }

src/rapaio/experiment/ml/classifier/meta/CStacking.java

@@ -53,7 +53,7 @@
  * <p>
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> on 9/30/15.
  */
-public class CStacking extends AbstractClassifierModel<CStacking, ClassifierResult<CStacking>> implements Printable {
+public class CStacking extends AbstractClassifierModel<CStacking, ClassifierResult> implements Printable {
 
     private static final long serialVersionUID = -9087871586729573030L;
 
@@ -155,7 +155,7 @@ protected PredSetup baseFit(Frame df, boolean withClasses, boolean withDistribut
     }
 
     @Override
-    protected ClassifierResult<CStacking> corePredict(Frame df, boolean withClasses, boolean withDistributions) {
+    protected ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDistributions) {
         return ClassifierResult.copy(this, df, withClasses, withDistributions, stacker.predict(df));
     }
 }

src/rapaio/experiment/ml/classifier/meta/CStepwiseSelection.java

@@ -53,7 +53,7 @@
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> on 11/11/15.
  */
 public class CStepwiseSelection
-        extends AbstractClassifierModel<CStepwiseSelection, ClassifierResult<CStepwiseSelection>> implements Printable {
+        extends AbstractClassifierModel<CStepwiseSelection, ClassifierResult> implements Printable {
 
     private static final long serialVersionUID = 2642562123626893974L;
     ClassifierModel best;
@@ -224,7 +224,7 @@ protected boolean coreFit(Frame df, Var weights) {
     }
 
     @Override
-    protected ClassifierResult<CStepwiseSelection> corePredict(Frame df, boolean withClasses, boolean withDistributions) {
+    protected ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDistributions) {
         return ClassifierResult.copy(this, df, withClasses, withDistributions, best.predict(df, withClasses, withDistributions));
     }
 

src/rapaio/experiment/ml/classifier/meta/SplitClassifierModel.java

@@ -47,7 +47,7 @@
  */
 @Deprecated
 public class SplitClassifierModel
-        extends AbstractClassifierModel<SplitClassifierModel, ClassifierResult<SplitClassifierModel>> implements Printable {
+        extends AbstractClassifierModel<SplitClassifierModel, ClassifierResult> implements Printable {
 
     private static final long serialVersionUID = 3332377951136731541L;
 
@@ -126,15 +126,15 @@ public boolean coreFit(Frame df, Var weights) {
     }
 
     @Override
-    public ClassifierResult<SplitClassifierModel> corePredict(Frame df, boolean withClasses, boolean withDensities) {
+    public ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDensities) {
 
-        ClassifierResult<SplitClassifierModel> pred = ClassifierResult.build(this, df, withClasses, withDensities);
+        ClassifierResult pred = ClassifierResult.build(this, df, withClasses, withDensities);
         df.stream().forEach(spot -> {
             for (Split split : splits) {
                 if (split.predicate.test(spot)) {
 
                     Frame f = MappedFrame.byRow(df, spot.row());
-                    ClassifierResult<ClassifierModel> p = split.classifierModel.predict(f, withClasses, withDensities);
+                    ClassifierResult p = split.classifierModel.predict(f, withClasses, withDensities);
 
                     if (withClasses) {
                         for (String targetVar : targetNames()) {

src/rapaio/experiment/ml/classifier/svm/BinarySMO.java

@@ -57,7 +57,7 @@
 /**
  * Class for building a binary support vector machine.
  */
-public class BinarySMO extends AbstractClassifierModel<BinarySMO, ClassifierResult<BinarySMO>> implements Serializable, Printable {
+public class BinarySMO extends AbstractClassifierModel<BinarySMO, ClassifierResult> implements Serializable, Printable {
 
     private static final long serialVersionUID = 1208515184777030598L;
 
@@ -442,8 +442,8 @@ protected boolean coreFit(Frame df, Var weights) {
 
 
     @Override
-    protected ClassifierResult<BinarySMO> corePredict(Frame df, boolean withClasses, boolean withDistributions) {
-        ClassifierResult<BinarySMO> cr = ClassifierResult.build(this, df, withClasses, withDistributions);
+    protected ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDistributions) {
+        ClassifierResult cr = ClassifierResult.build(this, df, withClasses, withDistributions);
         for (int i = 0; i < df.rowCount(); i++) {
             double pred = predict(df, i);
 

src/rapaio/experiment/ml/classifier/tree/CTree.java

@@ -68,7 +68,7 @@
  *
  * @author <a href="mailto:[email protected]">Aurelian Tutuianu</a>
  */
-public class CTree extends AbstractClassifierModel<CTree, ClassifierResult<CTree>> implements Printable {
+public class CTree extends AbstractClassifierModel<CTree, ClassifierResult> implements Printable {
 
     private static final long serialVersionUID = 1203926824359387358L;
     private static final Map<VType, CTreeTest> DEFAULT_TEST_MAP;
@@ -500,8 +500,8 @@ private void buildIndexMap(CTreeNode node, HashMap<Integer, Integer> indexMap) {
     }
 
     @Override
-    protected ClassifierResult<CTree> corePredict(Frame df, boolean withClasses, boolean withDensities) {
-        ClassifierResult<CTree> prediction = ClassifierResult.build(this, df, withClasses, withDensities);
+    protected ClassifierResult corePredict(Frame df, boolean withClasses, boolean withDensities) {
+        ClassifierResult prediction = ClassifierResult.build(this, df, withClasses, withDensities);
         for (int i = 0; i < df.rowCount(); i++) {
             Pair<String, DensityVector<String>> res = predictPoint(this, root, i, df);
             String label = res._1;

src/rapaio/experiment/ml/classifier/tree/CTreeTest.java

@@ -126,7 +126,7 @@ public CTreeCandidate computeCandidate(CTree c, Frame df, Var weights, String te
             int[] rows = new int[df.rowCount()];
             int len = 0;
             for (int i = 0; i < df.rowCount(); i++) {
-                if(!df.isMissing(i, testNameIndex)) {
+                if (!df.isMissing(i, testNameIndex)) {
                     rows[len++] = i;
                     dt.increment(1, dt.colIndex().getIndex(df, targetName, i), weights.getDouble(i));
                 }
@@ -250,7 +250,7 @@ public CTreeCandidate computeCandidate(CTree c, Frame df, Var weights, String te
 
             double[] rowCounts = counts.rowTotals();
             for (int i = 1; i < df.levels(testName).size(); i++) {
-                if (rowCounts[i] < c.minCount())
+                if (rowCounts[i - 1] < c.minCount())
                     continue;
 
                 String testLabel = df.rvar(testName).levels().get(i);

src/rapaio/experiment/ml/regression/boost/GBTRegressionModel.java

@@ -45,7 +45,6 @@
 import rapaio.ml.regression.RegressionResult;
 import rapaio.ml.regression.simple.L2RegressionModel;
 import rapaio.ml.regression.tree.RTree;
-import rapaio.printer.Printable;
 import rapaio.printer.Printer;
 import rapaio.printer.opt.POption;
 
@@ -61,8 +60,7 @@
  * User: Aurelian Tutuianu <[email protected]>
  */
 @Deprecated
-public class GBTRegressionModel extends AbstractRegressionModel<GBTRegressionModel, RegressionResult<GBTRegressionModel>>
-        implements Printable {
+public class GBTRegressionModel extends AbstractRegressionModel<GBTRegressionModel, RegressionResult> {
 
     private static final long serialVersionUID = 4559540258922653130L;
 
@@ -197,9 +195,9 @@ protected boolean coreFit(Frame df, Var weights) {
     }
 
     @Override
-    protected RegressionResult<GBTRegressionModel> corePredict(final Frame df, final boolean withResiduals) {
-        RegressionResult<GBTRegressionModel> pred = RegressionResult.build(this, df, withResiduals);
-        RegressionResult<GBTRegressionModel> initPred = initRegressionModel.predict(df, false);
+    protected RegressionResult corePredict(final Frame df, final boolean withResiduals) {
+        RegressionResult pred = RegressionResult.build(this, df, withResiduals);
+        RegressionResult initPred = initRegressionModel.predict(df, false);
         for (int i = 0; i < df.rowCount(); i++) {
             pred.firstPrediction().setDouble(i, initPred.firstPrediction().getDouble(i));
         }

src/rapaio/experiment/ml/regression/ensemble/RForest.java

@@ -37,7 +37,6 @@
 import rapaio.ml.regression.RegressionModel;
 import rapaio.ml.regression.RegressionResult;
 import rapaio.ml.regression.tree.RTree;
-import rapaio.printer.Printable;
 import rapaio.printer.Printer;
 import rapaio.printer.opt.POption;
 
@@ -50,8 +49,7 @@
 /**
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> at 1/15/15.
  */
-public class RForest extends AbstractRegressionModel<RForest, RegressionResult<RForest>>
-        implements Printable {
+public class RForest extends AbstractRegressionModel<RForest, RegressionResult> {
 
     private static final long serialVersionUID = -3926256335736143438L;
 
@@ -129,8 +127,8 @@ public List<RegressionModel> getRegressors() {
     }
 
     @Override
-    protected RegressionResult<RForest> corePredict(Frame df, boolean withResiduals) {
-        RegressionResult<RForest> fit = RegressionResult.build(this, df, withResiduals);
+    protected RegressionResult corePredict(Frame df, boolean withResiduals) {
+        RegressionResult fit = RegressionResult.build(this, df, withResiduals);
         List<VarDouble> results = regressors
                 .parallelStream()
                 .map(r -> r.predict(df, false).firstPrediction())

src/rapaio/experiment/ml/regression/tree/GBTRtree.java

@@ -36,7 +36,7 @@
 /**
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> on 5/21/19.
  */
-public interface GBTRtree<M extends RegressionModel<M, R>, R extends RegressionResult<M>> extends RegressionModel<M, R> {
+public interface GBTRtree<M extends RegressionModel, R extends RegressionResult> extends RegressionModel {
 
     void boostUpdate(Frame x, Var y, Var fx, GBTRegressionLoss lossFunction);
 }

src/rapaio/experiment/ml/regression/tree/NestedBoostingRTree.java

@@ -47,8 +47,8 @@
 /**
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> on 4/16/19.
  */
-public class NestedBoostingRTree extends AbstractRegressionModel<NestedBoostingRTree, RegressionResult<NestedBoostingRTree>>
-        implements GBTRtree<NestedBoostingRTree, RegressionResult<NestedBoostingRTree>> {
+public class NestedBoostingRTree extends AbstractRegressionModel<NestedBoostingRTree, RegressionResult>
+        implements GBTRtree<NestedBoostingRTree, RegressionResult> {
 
     private static final long serialVersionUID = 1864784340491461993L;
     private int minCount = 5;

src/rapaio/experiment/ml/regression/tree/SmoothRTree.java

@@ -48,8 +48,8 @@
 /**
  * Created by <a href="mailto:[email protected]">Aurelian Tutuianu</a> on 6/19/19.
  */
-public class SmoothRTree extends AbstractRegressionModel<SmoothRTree, RegressionResult<SmoothRTree>>
-        implements GBTRtree<SmoothRTree, RegressionResult<SmoothRTree>> {
+public class SmoothRTree extends AbstractRegressionModel<SmoothRTree, RegressionResult>
+        implements GBTRtree<SmoothRTree, RegressionResult> {
 
     private static final long serialVersionUID = 5062591010395009141L;
 
@@ -168,8 +168,8 @@ protected boolean coreFit(Frame df, Var weights) {
     }
 
     @Override
-    protected RegressionResult<SmoothRTree> corePredict(Frame df, boolean withResiduals) {
-        RegressionResult<SmoothRTree> prediction = RegressionResult.build(this, df, withResiduals);
+    protected RegressionResult corePredict(Frame df, boolean withResiduals) {
+        RegressionResult prediction = RegressionResult.build(this, df, withResiduals);
         for (int i = 0; i < df.rowCount(); i++) {
             prediction.firstPrediction().setDouble(i, root.predict(df, i, this, 1.0));
         }

src/rapaio/ml/classifier/AbstractClassifierModel.java

@@ -52,8 +52,7 @@
  *
  * @author <a href="mailto:[email protected]">Aurelian Tutuianu</a>
  */
-public abstract class AbstractClassifierModel<M extends ClassifierModel<M, R>, R extends ClassifierResult<M>>
-        implements ClassifierModel<M, R> {
+public abstract class AbstractClassifierModel<M extends ClassifierModel, R extends ClassifierResult> implements ClassifierModel {
 
     private static final long serialVersionUID = -6866948033065091047L;
 
@@ -122,18 +121,18 @@ public BiConsumer<M, Integer> runningHook() {
     }
 
     @Override
-    public M withRunningHook(BiConsumer<M, Integer> runningHook) {
-        this.runningHook = runningHook;
-        return (M) this;
+    public <T extends ClassifierModel> T withRunningHook(BiConsumer<? extends ClassifierModel, Integer> runningHook) {
+        this.runningHook = (BiConsumer<M, Integer>) runningHook;
+        return (T)this;
     }
 
     public BiFunction<M, Integer, Boolean> stoppingHook() {
         return stoppingHook;
     }
 
-    public M withStoppingHook(BiFunction<M, Integer, Boolean> stoppingHook) {
-        this.stoppingHook = stoppingHook;
-        return (M) this;
+    public <T extends ClassifierModel> T withStoppingHook(BiFunction<? extends ClassifierModel, Integer, Boolean> stoppingHook) {
+        this.stoppingHook = (BiFunction<M, Integer, Boolean>) stoppingHook;
+        return (T) this;
     }
 
     @Override