Refactored AllCategoricalSplit

src/mlpack/methods/decision_tree/all_categorical_split.hpp

@@ -47,23 +47,23 @@ class AllCategoricalSplit
    * @param weights Weights associated with labels.
    * @param minimumLeafSize Minimum number of points in a leaf node for
    *      splitting.
-   * @param classProbabilities Class probabilities vector, which may be filled
-   *      with split information a successful split.
+   * @param splitInfo Stores split information on a successful split.
    * @param minimumGainSplit Minimum  gain split.
    * @param aux Auxiliary split information, which may be modified on a
    *      successful split.
    */
-  template<bool UseWeights, typename VecType, typename WeightVecType>
+  template<bool UseWeights, typename VecType, typename ElemType, typename WeightVecType>
   static double SplitIfBetter(
       const double bestGain,
       const VecType& data,
       const size_t numCategories,
-      const arma::Row<size_t>& labels,
+      const arma::Row<ElemType>& labels,
+      const size_t begin,
       const size_t numClasses,
       const WeightVecType& weights,
       const size_t minimumLeafSize,
       const double minimumGainSplit,
-      arma::vec& classProbabilities,
+      double& splitInfo,
       AuxiliarySplitInfo& aux);
 
   /**

src/mlpack/methods/decision_tree/all_categorical_split_impl.hpp

@@ -16,17 +16,18 @@ namespace mlpack {
 namespace tree {
 
 template<typename FitnessFunction>
-template<bool UseWeights, typename VecType, typename WeightVecType>
+template<bool UseWeights, typename VecType, typename ElemType, typename WeightVecType>
 double AllCategoricalSplit<FitnessFunction>::SplitIfBetter(
     const double bestGain,
     const VecType& data,
     const size_t numCategories,
-    const arma::Row<size_t>& labels,
+    const arma::Row<ElemType>& labels,
+    const size_t begin,
     const size_t numClasses,
     const WeightVecType& weights,
     const size_t minimumLeafSize,
     const double minimumGainSplit,
-    arma::vec& classProbabilities,
+    double& splitInfo,
     AuxiliarySplitInfo& /* aux */)
 {
   // Count the number of elements in each potential child.
@@ -58,7 +59,7 @@ double AllCategoricalSplit<FitnessFunction>::SplitIfBetter(
   // Calculate the gain of the split.  First we have to calculate the labels
   // that would be assigned to each child.
   arma::uvec childPositions(numCategories, arma::fill::zeros);
-  std::vector<arma::Row<size_t>> childLabels(numCategories);
+  std::vector<arma::Row<ElemType>> childLabels(numCategories);
   std::vector<arma::Row<double>> childWeights(numCategories);
   for (size_t i = 0; i < numCategories; ++i)
   {
@@ -75,12 +76,12 @@ double AllCategoricalSplit<FitnessFunction>::SplitIfBetter(
 
     if (UseWeights)
     {
-      childLabels[category][childPositions[category]] = labels[i];
+      childLabels[category][childPositions[category]] = labels[begin + i];
       childWeights[category][childPositions[category]++] = weights[i];
     }
     else
     {
-      childLabels[category][childPositions[category]++] = labels[i];
+      childLabels[category][childPositions[category]++] = labels[begin + i];
     }
   }
 
@@ -99,9 +100,8 @@ double AllCategoricalSplit<FitnessFunction>::SplitIfBetter(
 
   if (overallGain > bestGain + minimumGainSplit + epsilon)
   {
-    // This is better, so set up the class probabilities vector and return.
-    classProbabilities.set_size(1);
-    classProbabilities[0] = numCategories;
+    // This is better, so store it in splitInfo and return.
+    splitInfo = numCategories;
     return overallGain;
   }
 

src/mlpack/methods/decision_tree/decision_tree_impl.hpp

@@ -642,15 +642,17 @@ double DecisionTree<FitnessFunction,
       double dimGain = -DBL_MAX;
       if (datasetInfo.Type(i) == data::Datatype::categorical)
       {
+        classProbabilities.set_size(1);
         dimGain = CategoricalSplit::template SplitIfBetter<UseWeights>(bestGain,
             data.cols(begin, begin + count - 1).row(i),
             datasetInfo.NumMappings(i),
-            labels.subvec(begin, begin + count - 1),
+            labels,
+            begin,
             numClasses,
             UseWeights ? weights.subvec(begin, begin + count - 1) : weights,
             minimumLeafSize,
             minimumGainSplit,
-            classProbabilities,
+            classProbabilities[0],
             *this);
       }
       else if (datasetInfo.Type(i) == data::Datatype::numeric)

src/mlpack/tests/decision_tree_test.cpp

@@ -583,17 +583,17 @@ TEST_CASE("AllCategoricalSplitSimpleSplitTest", "[DecisionTreeTest]")
   arma::rowvec weights(labels.n_elem);
   weights.ones();
 
-  arma::vec classProbabilities;
+  arma::vec classProbabilities(1);
   AllCategoricalSplit<GiniGain>::AuxiliarySplitInfo aux;
 
   // Call the method to do the splitting.
   const double bestGain = GiniGain::Evaluate<false>(labels, 3, weights);
   const double gain = AllCategoricalSplit<GiniGain>::SplitIfBetter<false>(
-      bestGain, values, 4, labels, 3, weights, 3, 1e-7, classProbabilities,
+      bestGain, values, 4, labels, 0, 3, weights, 3, 1e-7, classProbabilities[0],
       aux);
   const double weightedGain =
       AllCategoricalSplit<GiniGain>::SplitIfBetter<true>(bestGain, values, 4,
-      labels, 3, weights, 3, 1e-7, classProbabilities, aux);
+      labels, 0, 3, weights, 3, 1e-7, classProbabilities[0], aux);
 
   // Make sure that a split was made.
   REQUIRE(gain > bestGain);
@@ -619,18 +619,17 @@ TEST_CASE("AllCategoricalSplitMinSamplesTest", "[DecisionTreeTest]")
   arma::rowvec weights(labels.n_elem);
   weights.ones();
 
-  arma::vec classProbabilities;
+  arma::vec classProbabilities(1);
   AllCategoricalSplit<GiniGain>::AuxiliarySplitInfo aux;
 
   // Call the method to do the splitting.
   const double bestGain = GiniGain::Evaluate<false>(labels, 3, weights);
   const double gain = AllCategoricalSplit<GiniGain>::SplitIfBetter<false>(
-      bestGain, values, 4, labels, 3, weights, 4, 1e-7, classProbabilities,
-      aux);
+      bestGain, values, 4, labels, 0, 3, weights, 4, 1e-7,
+      classProbabilities[0], aux);
 
   // Make sure it's not split.
   REQUIRE(gain == DBL_MAX);
-  REQUIRE(classProbabilities.n_elem == 0);
 }
 
 /**
@@ -652,22 +651,21 @@ TEST_CASE("AllCategoricalSplitNoGainTest", "[DecisionTreeTest]")
     labels[i + 2] = 2;
   }
 
-  arma::vec classProbabilities;
+  arma::vec classProbabilities(1);
   AllCategoricalSplit<GiniGain>::AuxiliarySplitInfo aux;
 
   // Call the method to do the splitting.
   const double bestGain = GiniGain::Evaluate<false>(labels, 3, weights);
   const double gain = AllCategoricalSplit<GiniGain>::SplitIfBetter<false>(
-      bestGain, values, 10, labels, 3, weights, 10, 1e-7,
-      classProbabilities, aux);
+      bestGain, values, 10, labels, 0, 3, weights, 10, 1e-7,
+      classProbabilities[0], aux);
   const double weightedGain =
       AllCategoricalSplit<GiniGain>::SplitIfBetter<true>(bestGain, values, 10,
-      labels, 3, weights, 10, 1e-7, classProbabilities, aux);
+      labels, 0, 3, weights, 10, 1e-7, classProbabilities[0], aux);
 
   // Make sure that there was no split.
   REQUIRE(gain == DBL_MAX);
   REQUIRE(gain == weightedGain);
-  REQUIRE(classProbabilities.n_elem == 0);
 }
 
 /**