Fixed implementation bug and normal distribution test

src/mlpack/methods/decision_tree/mad_gain.hpp

@@ -109,6 +109,8 @@ class MADGain
 
       accWeights[0] += accWeights[1] + accWeights[2] + accWeights[3];
       weightedMean[0] += weightedMean[1] + weightedMean[2] + weightedMean[3];
+      weightedMean[0] /= (double) (end - begin);
+      std::cout << "WeightedMean: " << weightedMean[0] << std::endl;
 
       // Catch edge case: if there are no weights, the impurity is zero.
       if (accWeights[0] == 0.0)
@@ -152,6 +154,8 @@ class MADGain
       }
 
       mean[0] += mean[1] + mean[2] + mean[3];
+      mean[0] /= (double) (end - begin);
+      std::cout << "Mean: " << mean[0] << std::endl;
 
       for (size_t i = begin; i < end; ++i)
         mad += std::abs(labels[i] - mean[0]);

src/mlpack/tests/decision_tree_test.cpp

@@ -39,13 +39,13 @@ TEST_CASE("MADGainPerfectTest", "[DecisionTreeRegressionTest]")
 }
 
 /**
- * Make sure that for a normal distribution of labels,
- * MAD_gain = mean of absolute values of the distribution.
+ * Make sure that when mean of labels is zero, MAD_gain = mean of
+ * absolute values of the distribution.
  */
 TEST_CASE("MADGainNormalTest", "[DecisionTreeRegressionTest")
 {
   arma::rowvec weights(10, arma::fill::ones);
-  arma::rowvec labels(10, arma::fill::randn); // Mean = 0.
+  arma::rowvec labels = { 1, 2, 3, 4, 5, -1, -2, -3, -4, -5 }; // Mean = 0.
 
   // Theoretical gain.
   double theoreticalGain = 0.0;