Move SplitData to data namespace.

It seems more appropriate to me; the util namespace primarily contains utilities
for C++ programming in mlpack, whereas the data namespace contains utilities for
manipulating data.

src/mlpack/core/data/split_data.hpp

@@ -4,7 +4,7 @@
 #include <mlpack/core.hpp>
 
 namespace mlpack {
-namespace util {
+namespace data {
 
 /**
  *Split training data and test data
@@ -95,7 +95,7 @@ TrainTestSplit(const arma::Mat<T> &input,
                          trainLabel, testLabel);
 }
 
-} // namespace util
+} // namespace data
 } // namespace mlpack
 
 #endif

src/mlpack/tests/split_data_test.cpp

@@ -4,15 +4,15 @@
  *
  * Test the SparseAutoencoder class.
  */
-
 #include <mlpack/core.hpp>
-#include <mlpack/core/util/split_data.hpp>
+#include <mlpack/core/data/split_data.hpp>
 
 #include <boost/test/unit_test.hpp>
 #include "old_boost_test_definitions.hpp"
 
 using namespace mlpack;
 using namespace arma;
+using namespace mlpack::data;
 
 BOOST_AUTO_TEST_SUITE(SplitDataTest);
 
@@ -46,7 +46,7 @@ BOOST_AUTO_TEST_CASE(SplitDataSplitResultMat)
           arma::linspace<Labels>(0, input.n_cols-1,
                                  input.n_cols);
 
-  auto const value = util::TrainTestSplit(input, labels, 0.2);
+  auto const value = TrainTestSplit(input, labels, 0.2);
   BOOST_REQUIRE(std::get<0>(value).n_cols == 8);
   BOOST_REQUIRE(std::get<1>(value).n_cols == 2);
   BOOST_REQUIRE(std::get<2>(value).n_cols == 8);