PUBDEV-4940 implement UpliftRandomForest

h2o-algos/src/main/java/hex/DataInfo.java

@@ -19,7 +19,7 @@
  * Supports sparse data, sparse columns can be transformed to sparse rows on the fly with some (significant) memory overhead,
  * as the data of the whole chunk(s) will be copied.
  *
-*/
+ */
 public class DataInfo extends Keyed<DataInfo> {
   public int [] _activeCols;
   public Frame _adaptedFrame;  // the modified DataInfo frame (columns sorted by largest categorical -> least then all numerical columns)
@@ -65,7 +65,6 @@ public void addResponse(String [] names, Vec[] vecs) {
 
   public double[] numNAFill() {return _numNAFill; }
   public double numNAFill(int nid) {return _numNAFill[nid];}
-
   public void setCatNAFill(int[] catNAFill) {
     _catNAFill = catNAFill;
   }
@@ -129,17 +128,18 @@ public boolean isSigmaScaled(){
   public boolean _offset;
   public boolean _weights;
   public boolean _fold;
+  public boolean _treatment;
   public Model.InteractionPair[] _interactions; // raw set of interactions
   public Model.InteractionSpec _interactionSpec; // formal specification of interactions
   public int _interactionVecs[]; // the interaction columns appearing in _adaptedFrame
   public int[] _numOffsets; // offset column indices used by numerical interactions: total number of numerical columns is given by _numOffsets[_nums] - _numOffsets[0]
-  public int responseChunkId(int n){return n + _cats + _nums + (_weights?1:0) + (_offset?1:0) + (_fold?1:0);}
+  public int responseChunkId(int n){return n + _cats + _nums + (_weights?1:0) + (_offset?1:0) + (_fold?1:0) + (_treatment?1:0);}
+  public int treatmentChunkId(){return _cats + _nums + (_weights?1:0) + (_offset?1:0) + (_fold?1:0);}
   public int foldChunkId(){return _cats + _nums + (_weights?1:0) + (_offset?1:0);}
-
   public int offsetChunkId(){return _cats + _nums + (_weights ?1:0);}
   public int weightChunkId(){return _cats + _nums;}
   public int outputChunkId() { return outputChunkId(0);}
-  public int outputChunkId(int n) { return n + _cats + _nums + (_weights?1:0) + (_offset?1:0) + (_fold?1:0) + _responses;}
+  public int outputChunkId(int n) { return n + _cats + _nums + (_weights?1:0) + (_offset?1:0) + (_fold?1:0) + (_treatment?1:0) + _responses;}
   public void addOutput(String name, Vec v) {_adaptedFrame.add(name,v);}
   public Vec getOutputVec(int i) {return _adaptedFrame.vec(outputChunkId(i));}
   public void setResponse(String name, Vec v){ setResponse(name,v,0);}
@@ -151,7 +151,7 @@ public boolean isSigmaScaled(){
   public final int [][] _catLvls; // cat lvls post filter (e.g. by strong rules)
   public final int [][] _intLvls; // interaction lvls post filter (e.g. by strong rules)
 
-  private DataInfo() {  _intLvls=null; _catLvls = null; _skipMissing = true; _imputeMissing = false; _valid = false; _offset = false; _weights = false; _fold = false; }
+  private DataInfo() {  _intLvls=null; _catLvls = null; _skipMissing = true; _imputeMissing = false; _valid = false; _offset = false; _weights = false; _fold = false; _treatment=false;}
   public String[] _coefNames;
   public int[] _coefOriginalIndices; // 
   @Override protected long checksum_impl() {throw H2O.unimpl();} // don't really need checksum
@@ -168,7 +168,15 @@ public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLe
   }
 
   public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLevels, TransformType predictor_transform, TransformType response_transform, boolean skipMissing, boolean imputeMissing, boolean missingBucket, boolean weight, boolean offset, boolean fold, Model.InteractionSpec interactions) {
-    this(train, valid, nResponses, useAllFactorLevels, predictor_transform, response_transform, skipMissing, imputeMissing, new MeanImputer(), missingBucket, weight, offset, fold, interactions);
+    this(train, valid, nResponses, useAllFactorLevels, predictor_transform, response_transform, skipMissing, imputeMissing, new MeanImputer(), missingBucket, weight, offset, fold, false, interactions);
+  }
+
+  public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLevels, TransformType predictor_transform, TransformType response_transform, boolean skipMissing, boolean imputeMissing, boolean missingBucket, boolean weight, boolean offset, boolean fold, boolean treatment, Model.InteractionSpec interactions) {
+    this(train, valid, nResponses, useAllFactorLevels, predictor_transform, response_transform, skipMissing, imputeMissing, new MeanImputer(), missingBucket, weight, offset, fold, treatment, interactions);
+  }
+
+  public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLevels, TransformType predictor_transform, TransformType response_transform, boolean skipMissing, boolean imputeMissing, Imputer imputer, boolean missingBucket, boolean weight, boolean offset, boolean fold, Model.InteractionSpec interactions) {
+    this(train, valid, nResponses, useAllFactorLevels, predictor_transform, response_transform, skipMissing, imputeMissing, imputer, missingBucket, weight, offset, fold, false, interactions);
   }
 
   /**
@@ -191,14 +199,15 @@ public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLe
    *    A. As a list of pairs of column indices.
    *    B. As a list of pairs of column indices with limited enums.
    */
-  public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLevels, TransformType predictor_transform, TransformType response_transform, boolean skipMissing, boolean imputeMissing, Imputer imputer, boolean missingBucket, boolean weight, boolean offset, boolean fold, Model.InteractionSpec interactions) {
+  public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLevels, TransformType predictor_transform, TransformType response_transform, boolean skipMissing, boolean imputeMissing, Imputer imputer, boolean missingBucket, boolean weight, boolean offset, boolean fold, boolean treatment, Model.InteractionSpec interactions) {
     super(Key.<DataInfo>make());
     assert predictor_transform != null;
     assert response_transform != null;
     _valid = valid != null;
     _offset = offset;
     _weights = weight;
     _fold = fold;
+    _treatment = treatment;
     assert !(skipMissing && imputeMissing) : "skipMissing and imputeMissing cannot both be true";
     _skipMissing = skipMissing;
     _imputeMissing = imputeMissing;
@@ -228,7 +237,7 @@ public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLe
     final Vec[] tvecs = train.vecs();
 
     // Count categorical-vs-numerical
-    final int n = tvecs.length-_responses - (offset?1:0) - (weight?1:0) - (fold?1:0);
+    final int n = tvecs.length-_responses - (offset?1:0) - (weight?1:0) - (fold?1:0) - (treatment?1:0);
     int [] nums = MemoryManager.malloc4(n);
     int [] cats = MemoryManager.malloc4(n);
     int nnums = 0, ncats = 0;
@@ -323,7 +332,7 @@ public DataInfo(Frame train, Frame valid, int nResponses, boolean useAllFactorLe
         numIdx++;
       }
     }
-    for(int i = names.length-nResponses - (weight?1:0) - (offset?1:0) - (fold?1:0); i < names.length; ++i) {
+    for(int i = names.length-nResponses - (weight?1:0) - (offset?1:0) - (fold?1:0) - (treatment?1:0); i < names.length; ++i) {
       names[i] = train._names[i];
       tvecs2[i] = train.vec(i);
     }
@@ -423,6 +432,7 @@ private DataInfo(DataInfo dinfo,Frame fr, double [] normMul, double [] normSub,
     _offset = dinfo._offset;
     _weights = dinfo._weights;
     _fold = dinfo._fold;
+    _treatment = dinfo._treatment;
     _valid = false;
     _interactions = null;
     ArrayList<Integer> interactionVecs = new ArrayList<>();
@@ -588,7 +598,7 @@ public DataInfo filterExpandedColumns(int [] cols){
         normMul[k-id] = _normMul[cols[k]-off];
     }
     DataInfo dinfo = new DataInfo(this,f,normMul,normSub,catLvls,intLvls,catModes,cols);
-    dinfo._nums=f.numCols()-dinfo._cats - dinfo._responses - (dinfo._offset?1:0) - (dinfo._weights?1:0) - (dinfo._fold?1:0);
+    dinfo._nums=f.numCols()-dinfo._cats - dinfo._responses - (dinfo._offset?1:0) - (dinfo._weights?1:0) - (dinfo._fold?1:0) - (dinfo._treatment?1:0);
     dinfo._numMeans=new double[nnums];
     dinfo._numNAFill=new double[nnums];
     int colsSize = id+nnums;  // small optimization
@@ -735,6 +745,7 @@ public boolean isInteractionVec(int colid) {
    *  weight column
    *  offset column
    *  fold column
+   *  treatment column
    *
    * @return expanded number of columns in the underlying frame
    */
@@ -1450,6 +1461,7 @@ public DataInfo scoringInfo(String[] names, Frame adaptFrame, int nResponses, bo
     res._weights = _weights && adaptFrame.find(names[weightChunkId()]) != -1;
     res._offset = _offset && adaptFrame.find(names[offsetChunkId()]) != -1;
     res._fold = _fold && adaptFrame.find(names[foldChunkId()]) != -1;
+    res._treatment = _treatment && adaptFrame.find(names[treatmentChunkId()]) != -1;
     if (nResponses != -1) {
       res._responses = nResponses;
     } else {

h2o-algos/src/main/java/hex/api/RegisterAlgos.java

@@ -34,6 +34,7 @@ public void registerEndPoints(RestApiContext context) {
             new ANOVAGLM(true),
             new PSVM(true),
             new hex.rulefit     .RuleFit     (true),
+            new hex.tree.uplift.UpliftDRF    (true),
             new hex.maxrglm     .MaxRGLM     (true)
     };
 

h2o-algos/src/main/java/hex/generic/GenericModel.java

@@ -143,6 +143,8 @@ public String getResponseColumn() {
                 return genModel.isSupervised() ? genModel.getResponseName() : null; 
             }
             @Override
+            public String getTreatmentColumn() {return null;}
+            @Override
             public double missingColumnsType() {
                 return Double.NaN;
             }

h2o-algos/src/main/java/hex/schemas/UpliftDRFModelV3.java

@@ -0,0 +1,26 @@
+package hex.schemas;
+
+import hex.tree.uplift.UpliftDRFModel;
+
+public class UpliftDRFModelV3 extends SharedTreeModelV3<UpliftDRFModel,
+            UpliftDRFModelV3,
+            UpliftDRFModel.UpliftDRFParameters,
+            UpliftDRFV3.UpliftDRFParametersV3,
+            UpliftDRFModel.UpliftDRFOutput,
+            UpliftDRFModelV3.UpliftDRFModelOutputV3> {
+
+        public static final class UpliftDRFModelOutputV3 extends SharedTreeModelV3.SharedTreeModelOutputV3<UpliftDRFModel.UpliftDRFOutput, UpliftDRFModelOutputV3> {}
+
+        public UpliftDRFV3.UpliftDRFParametersV3 createParametersSchema() { return new UpliftDRFV3.UpliftDRFParametersV3(); }
+        public UpliftDRFModelOutputV3 createOutputSchema() { return new UpliftDRFModelOutputV3(); }
+
+        //==========================
+        // Custom adapters go here
+
+        // Version&Schema-specific filling into the impl
+        @Override public UpliftDRFModel createImpl() {
+            UpliftDRFV3.UpliftDRFParametersV3 p = this.parameters;
+            UpliftDRFModel.UpliftDRFParameters parms = p.createImpl();
+            return new UpliftDRFModel( model_id.key(), parms, new UpliftDRFModel.UpliftDRFOutput(null) );
+        }
+}

h2o-algos/src/main/java/hex/schemas/UpliftDRFV3.java

@@ -0,0 +1,73 @@
+package hex.schemas;
+
+import hex.AUUC;
+import hex.tree.uplift.UpliftDRF;
+import hex.tree.uplift.UpliftDRFModel.UpliftDRFParameters;
+import water.api.API;
+
+public class UpliftDRFV3 extends SharedTreeV3<UpliftDRF, UpliftDRFV3, UpliftDRFV3.UpliftDRFParametersV3> {
+
+    public static final class UpliftDRFParametersV3 extends SharedTreeV3.SharedTreeParametersV3<UpliftDRFParameters, UpliftDRFParametersV3> {
+        static public String[] fields = new String[]{
+                "model_id",
+                "training_frame",
+                "validation_frame",
+                "score_each_iteration",
+                "score_tree_interval",
+                "response_column",
+                "ignored_columns",
+                "ignore_const_cols",
+                "balance_classes",
+                "class_sampling_factors",
+                "max_after_balance_size",
+                "ntrees",
+                "max_depth",
+                "min_rows",
+                "nbins",
+                "nbins_top_level",
+                "nbins_cats",
+                "max_runtime_secs",
+                "seed",
+                "mtries",
+                "sample_rate",
+                "sample_rate_per_class",
+                "checkpoint",
+                "col_sample_rate_change_per_level",
+                "col_sample_rate_per_tree",
+                "histogram_type",
+                "categorical_encoding",
+                "calibrate_model",
+                "calibration_frame",
+                "distribution",
+                "custom_metric_func",
+                "export_checkpoints_dir",
+                "check_constant_response",
+                "treatment_column",
+                "uplift_metric",
+                "auuc_type",
+                "auuc_nbins"
+        };
+
+        // Input fields
+        @API(help = "Number of variables randomly sampled as candidates at each split. If set to -1, defaults to sqrt{p} for classification and p/3 for regression (where p is the # of predictors", gridable = true)
+        public int mtries;
+
+        @API(help = "Row sample rate per tree (from 0.0 to 1.0)", gridable = true)
+        public double sample_rate;
+
+        @API(help = "Define column which will be use for computing uplift gain to select best split for a tree. The column has to devide dataset into treatment (value 1) and control (value 0) group.", gridable = false, level = API.Level.secondary, required = true,
+                is_member_of_frames = {"training_frame", "validation_frame"},
+                is_mutually_exclusive_with = {"ignored_columns","response_column", "weights_column"})
+        public String treatment_column;
+
+        @API(help = "Divergence metric used to find best split when building an upplift tree.", level = API.Level.secondary, values = { "AUTO", "KL", "Euclidean", "ChiSquared"})
+        public UpliftDRFParameters.UpliftMetricType uplift_metric;
+
+        @API(help = "AUUC metric used to calculate Area under Uplift.", level = API.Level.secondary, values = { "AUTO", "Qini", "Lift", "Gain"})
+        public AUUC.AUUCType auuc_type;
+
+        @API(help = "Number of bins to calculate Area under Uplift.", level = API.Level.secondary)
+        public int auuc_nbins;
+
+    }
+}