[mlpack-git] master: Fix compilation errors; add a first test. (06eff7a)

[gitdub at big.cc.gt.atl.ga.us]

Repository : https://github.com/mlpack/mlpack

On branch  : master
Link       : https://github.com/mlpack/mlpack/compare/de9cc4b05069e1fa4793d9355f2f595af5ff45d2...6070527af14296cd99739de6c62666cc5d2a2125

>---------------------------------------------------------------

commit 06eff7a081d51b91e4a02ffd9176c42499ab614e
Author: ryan <ryan at ratml.org>
Date:   Thu Oct 8 11:23:52 2015 -0400

    Fix compilation errors; add a first test.


>---------------------------------------------------------------

06eff7a081d51b91e4a02ffd9176c42499ab614e
 src/mlpack/methods/hoeffding_trees/CMakeLists.txt  |  2 ++
 .../hoeffding_trees/binary_numeric_split.hpp       |  4 ++--
 .../hoeffding_trees/binary_numeric_split_impl.hpp  | 10 +++++-----
 src/mlpack/tests/hoeffding_tree_test.cpp           | 23 ++++++++++++++++++++++
 4 files changed, 32 insertions(+), 7 deletions(-)

diff --git a/src/mlpack/methods/hoeffding_trees/CMakeLists.txt b/src/mlpack/methods/hoeffding_trees/CMakeLists.txt
index 957e609..9b79011 100644
--- a/src/mlpack/methods/hoeffding_trees/CMakeLists.txt
+++ b/src/mlpack/methods/hoeffding_trees/CMakeLists.txt
@@ -1,6 +1,8 @@
 # Define the files we need to compile
 # Anything not in this list will not be compiled into MLPACK.
 set(SOURCES
+  binary_numeric_split.hpp
+  binary_numeric_split_impl.hpp
   categorical_split_info.hpp
   gini_impurity.hpp
   hoeffding_categorical_split.hpp
diff --git a/src/mlpack/methods/hoeffding_trees/binary_numeric_split.hpp b/src/mlpack/methods/hoeffding_trees/binary_numeric_split.hpp
index d2d7662..5873594 100644
--- a/src/mlpack/methods/hoeffding_trees/binary_numeric_split.hpp
+++ b/src/mlpack/methods/hoeffding_trees/binary_numeric_split.hpp
@@ -43,9 +43,9 @@ class BinaryNumericSplit
 
   void Train(ObservationType value, const size_t label);
 
-  double EvaluateFitnessFunction() const;
+  double EvaluateFitnessFunction();
 
-  void Split(arma::Col<size_t>& childMajorities, SplitInfo& splitInfo) const;
+  void Split(arma::Col<size_t>& childMajorities, SplitInfo& splitInfo);
 
   size_t MajorityClass() const;
   double MajorityProbability() const;
diff --git a/src/mlpack/methods/hoeffding_trees/binary_numeric_split_impl.hpp b/src/mlpack/methods/hoeffding_trees/binary_numeric_split_impl.hpp
index 15da9b2..1c45829 100644
--- a/src/mlpack/methods/hoeffding_trees/binary_numeric_split_impl.hpp
+++ b/src/mlpack/methods/hoeffding_trees/binary_numeric_split_impl.hpp
@@ -39,7 +39,7 @@ void BinaryNumericSplit<FitnessFunction, ObservationType>::Train(
 
 template<typename FitnessFunction, typename ObservationType>
 double BinaryNumericSplit<FitnessFunction, ObservationType>::
-    EvaluateFitnessFunction() const
+    EvaluateFitnessFunction()
 {
   // Unfortunately, we have to iterate over the map.
   bestSplit = std::numeric_limits<ObservationType>::min();
@@ -51,7 +51,7 @@ double BinaryNumericSplit<FitnessFunction, ObservationType>::
 
   double bestValue = FitnessFunction::Evaluate(counts);
 
-  for (std::multimap<ObservationType, size_t>::const_iterator it =
+  for (typename std::multimap<ObservationType, size_t>::const_iterator it =
       sortedElements.begin(); it != sortedElements.end(); ++it)
   {
     // Move the point to the right side of the split.
@@ -75,7 +75,7 @@ double BinaryNumericSplit<FitnessFunction, ObservationType>::
 template<typename FitnessFunction, typename ObservationType>
 void BinaryNumericSplit<FitnessFunction, ObservationType>::Split(
     arma::Col<size_t>& childMajorities,
-    SplitInfo& splitInfo) const
+    SplitInfo& splitInfo)
 {
   if (!isAccurate)
     EvaluateFitnessFunction();
@@ -87,8 +87,8 @@ void BinaryNumericSplit<FitnessFunction, ObservationType>::Split(
   counts.col(0).zeros();
   counts.col(1) = classCounts;
 
-  for (std::multimap<ObservationType, size_t>::const_iterator it =
-      sortedElements.begin(); (*it).second <= bestValue; ++it)
+  for (typename std::multimap<ObservationType, size_t>::const_iterator it =
+      sortedElements.begin(); (*it).second <= bestSplit; ++it)
   {
     // Move the point to the correct side of the split.
     --counts((*it).second, 1);
diff --git a/src/mlpack/tests/hoeffding_tree_test.cpp b/src/mlpack/tests/hoeffding_tree_test.cpp
index 6d628c5..3267feb 100644
--- a/src/mlpack/tests/hoeffding_tree_test.cpp
+++ b/src/mlpack/tests/hoeffding_tree_test.cpp
@@ -9,6 +9,7 @@
 #include <mlpack/methods/hoeffding_trees/gini_impurity.hpp>
 #include <mlpack/methods/hoeffding_trees/hoeffding_split.hpp>
 #include <mlpack/methods/hoeffding_trees/hoeffding_categorical_split.hpp>
+#include <mlpack/methods/hoeffding_trees/binary_numeric_split.hpp>
 
 #include <boost/test/unit_test.hpp>
 #include "old_boost_test_definitions.hpp"
@@ -523,4 +524,26 @@ BOOST_AUTO_TEST_CASE(HoeffdingNumericSplitBimodalTest)
   }
 }
 
+/**
+ * Create a BinaryNumericSplit object, feed it a bunch of samples where anything
+ * less than 1.0 is class 0 and anything greater is class 1.  Then make sure it
+ * can perform a perfect split.
+ */
+BOOST_AUTO_TEST_CASE(BinaryNumericSplitSimpleSplitTest)
+{
+  BinaryNumericSplit<GiniImpurity> split(2); // 2 classes.
+
+  // Feed it samples.
+  for (size_t i = 0; i < 500; ++i)
+  {
+    split.Train(mlpack::math::Random(), 0);
+    split.Train(mlpack::math::Random() + 1.0, 1);
+
+    // Now ensure the fitness function gives good gain.
+    // The Gini impurity for the unsplit node is 2 * (0.5^2) = 0.5, and the Gini
+    // impurity for the children is 0.
+    BOOST_REQUIRE_CLOSE(split.EvaluateFitnessFunction(), 0.5, 1e-5);
+  }
+}
+
 BOOST_AUTO_TEST_SUITE_END();