Merge branch 'master' of https://github.com/0xdata/h2o

a-b · Aug 8, 2013 · 03703bb · 03703bb
2 parents cf471c0 + f771da2
commit 03703bb
Show file tree

Hide file tree

Showing 4 changed files with 39 additions and 3 deletions.
diff --git a/src/main/java/hex/gbm/DHistogram.java b/src/main/java/hex/gbm/DHistogram.java
@@ -310,4 +310,22 @@ public void add( DHistogram h ) {
     }
     return sb.toString();
   }
+
+  private int byteSize(double[]ds) { return ds==null ? 0 : 24+ds.length*8; }
+  private int byteSize(long  []ls) { return ls==null ? 0 : 24+ls.length*8; }
+  private int byteSize(short []ss) { return ss==null ? 0 : 20+ss.length*2; }
+  public long byteSize() {
+    int sum = 14*8;             // Rough size of empty Histogram
+    sum += byteSize(_bins);
+    sum += byteSize(_mins);
+    sum += byteSize(_maxs);
+    sum += byteSize(_Ms);
+    sum += byteSize(_Ss);
+    sum += byteSize(_MSEs);
+    if( _clss != null ) {
+      sum += (_clss.length+3)*8;
+      for( long[] ls : _clss ) sum += byteSize(ls);
+    }
+    return sum;
+  }
 }
diff --git a/src/main/java/hex/gbm/DRF.java b/src/main/java/hex/gbm/DRF.java
@@ -44,7 +44,7 @@ private void run(Frame fr, int maxDepth, int ntrees, int mtrys, double sampleRat
     Timer t_drf = new Timer();
     assert 0 <= ntrees && ntrees < 1000000;
     assert 0 <= mtrys && mtrys < fr.numCols();
-    assert 0.0 <= sampleRate && sampleRate <= 1.0;
+    assert 0.0 < sampleRate && sampleRate <= 1.0;
 
     final String names[] = fr._names;
     Vec vs[] = fr._vecs;
@@ -120,8 +120,14 @@ public int makeSomeTrees( int st, DRFTree trees[], int leafs[], int ntrees, int
       for( int t=0; t<ntrees; t++ ) {
         final int tmax = trees[t]._len; // Number of total splits
         final DTree tree = trees[t];
-        for( int i=leafs[t]; i<tmax; i++ )
+        //int sum=0;
+        for( int i=leafs[t]; i<tmax; i++ ) {
           tree.undecided(i)._hs = sbh.getFinalHisto(t,i);
+          //Histogram hs[] = tree.undecided(i)._hs;
+          //for( Histogram h : hs )
+          //  if( h != null ) sum += h.byteSize();
+        }
+        //System.out.println("Tree#"+(st+t)+", leaves="+(trees[t]._len-leafs[t])+", histo size="+PrettyPrint.bytes(sum));
       }
 
       // Build up the next-generation tree splits from the current histograms.

diff --git a/src/main/java/water/AutoBuffer.java b/src/main/java/water/AutoBuffer.java
@@ -792,6 +792,12 @@ public byte[][] getAA1( ) {
     for( int i=0; i<len; i++ ) ary[i] = getA1();
     return ary;
   }
+  public short[][] getAA2( ) {
+    int len = get4();  if( len == -1 ) return null;
+    short[][] ary  = new short[len][];
+    for( int i=0; i<len; i++ ) ary[i] = getA2();
+    return ary;
+  }
   public int[][] getAA4( ) {
     int len = get4();  if( len == -1 ) return null;
     int[][] ary  = new int[len][];
@@ -906,6 +912,12 @@ public AutoBuffer putAA1( byte[][] ary ) {
     for( int i=0; i<ary.length; i++ ) putA1(ary[i]);
     return this;
   }
+  public AutoBuffer putAA2( short[][] ary ) {
+    if( ary == null ) return put4(-1);
+    put4(ary.length);
+    for( int i=0; i<ary.length; i++ ) putA2(ary[i]);
+    return this;
+  }
   public AutoBuffer putAA4( int[][] ary ) {
     if( ary == null ) return put4(-1);
     put4(ary.length);

diff --git a/src/main/java/water/api/DRF2.java b/src/main/java/water/api/DRF2.java
@@ -34,7 +34,7 @@ public class DRF2 extends Request {
   protected final FrameNonClassVecSelect vecs  = new FrameNonClassVecSelect("vecs",data_key,class_vec);
 
   @API(help="Sampling rate during tree building.")
-  protected final Real sample_rate = new Real("sample_rate", 0.67, 0.0, 1.0,"");
+  protected final Real sample_rate = new Real("sample_rate", 0.67, 0.000001, 1.0,"");
 
   @API(help="Psuedo-random number generator seed.")
   protected final LongInt seed = new LongInt("seed",0xae44a87f9edf1cbL,"High order bits make better seeds");