fix _pickle_method for predictor

.gitignore

@@ -23,3 +23,4 @@ old_data/*
 t2.py
 venv/*
 xgboost/*
+.idea/

auto_ml/DataFrameVectorizer.py

@@ -224,7 +224,7 @@ def transform_categorical_col(self, col_vals, col_name):
             encoded_col_names = []
 
             for trained_feature, col_idx in self.vocabulary_.items():
-                if trained_feature[:len_col_name] == col_name:
+                if trained_feature[:len_col_name] == col_name and '=' in trained_feature:
                     encoded_col_names.append([trained_feature, col_idx])
                     num_trained_cols += 1
                     if min_transformed_idx is None:

auto_ml/predictor.py

@@ -45,18 +45,18 @@
 from evolutionary_search import EvolutionaryAlgorithmSearchCV
 
 # For handling parallelism edge cases
-def _pickle_method(m):
-    if m.im_self is None:
-        return getattr, (m.im_class, m.im_func.func_name)
-    else:
-        return getattr, (m.im_self, m.im_func.func_name)
-
-try:
-    import copy_reg
+if sys.version_info[0] < 3:
+    import copy_reg as copy_reg
+
+
+    def _pickle_method(m):
+        if m.im_self is None:
+            return getattr, (m.im_class, m.im_func.func_name)
+        else:
+            return getattr, (m.im_self, m.im_func.func_name)
+
+
     copy_reg.pickle(types.MethodType, _pickle_method)
-except:
-    import copyreg
-    copyreg.pickle(types.MethodType, _pickle_method)
 
 
 class Predictor(object):

auto_ml/utils_models.py

@@ -591,7 +591,7 @@ def insert_deep_learning_model(pipeline_step, file_name):
     # Load the Keras model here
     keras_file_name = file_name[:-5] + random_name + '_keras_deep_learning_model.h5'
 
-    model = keras_load_model(keras_file_name)
+    model = load_keras_model(keras_file_name)
 
     # Put the model back in place so that we can still use it to get predictions without having to load it back in from disk
     return model