Added more docs and tests.

dlib/random_forest/random_forest_regression.h

@@ -46,6 +46,7 @@ namespace dlib
             std::vector<feature>& feats
         ) const
         {
+            DLIB_ASSERT(max_num_feats() != 0);
             num = std::min(num, num_feats);
 
             feats.clear();
@@ -66,6 +67,7 @@ namespace dlib
             const feature& f
         ) const
         {
+            DLIB_ASSERT(max_num_feats() != 0);
             return item(f);
         }
 
@@ -290,6 +292,7 @@ namespace dlib
             size_t num
         )
         {
+            DLIB_CASSERT(num > 0);
             num_trees = num;
         }
 
@@ -297,6 +300,7 @@ namespace dlib
             double frac
         )
         {
+            DLIB_CASSERT(0 < frac && frac <= 1);
             feature_subsampling_frac = frac;
         }
 
@@ -344,7 +348,7 @@ namespace dlib
         trained_function_type train (
             const std::vector<sample_type>& x,
             const std::vector<double>& y,
-            std::vector<double>& oob_values // predicted y, basically like LOO-CV
+            std::vector<double>& oob_values 
         ) const
         {
             return do_train(x,y,oob_values,true);
@@ -355,7 +359,7 @@ namespace dlib
         trained_function_type do_train (
             const std::vector<sample_type>& x,
             const std::vector<double>& y,
-            std::vector<double>& oob_values, // predicted y, basically like LOO-CV
+            std::vector<double>& oob_values, 
             bool compute_oob_values
         ) const 
         {

dlib/test/random_forest.cpp

@@ -41,6 +41,8 @@ namespace
         {
             istringstream sin(get_decoded_string());
 
+            print_spinner();
+
             typedef matrix<double,0,1> sample_type;
             std::vector<double> labels;
             std::vector<sample_type> samples;
@@ -61,7 +63,7 @@ namespace
 
             auto result = test_regression_function(df, samples, labels);
             // train:    2.239 0.987173 0.970669   1.1399 
-            dlog << LINFO << "train: " << trans(result);
+            dlog << LINFO << "train: " << result;
             DLIB_TEST_MSG(result(0) < 2.3, result(0));
 
             running_stats<double> rs;
@@ -69,6 +71,18 @@ namespace
                 rs.add(std::pow(oobs[i]-labels[i],2.0));
             dlog << LINFO << "OOB MSE: "<< rs.mean();
             DLIB_TEST_MSG(rs.mean() < 10.2, rs.mean());
+
+            print_spinner();
+
+            stringstream ss;
+            serialize(df, ss);
+            decltype(df) df2;
+            deserialize(df2, ss);
+            DLIB_TEST(df2.get_num_trees() == 1000);
+            result = test_regression_function(df2, samples, labels);
+            // train:    2.239 0.987173 0.970669   1.1399 
+            dlog << LINFO << "serialized train results: " << result;
+            DLIB_TEST_MSG(result(0) < 2.3, result(0));
         }
     } a;