changed defaults and added tests

rdt/transformers/datetime.py

@@ -23,7 +23,7 @@ class UnixTimestampEncoder(BaseTransformer):
             Indicate what to replace the null values with. If the strings ``'mean'`` or ``'mode'``
             are given, replace them with the corresponding aggregation, if ``'random'``, use
             random values from the dataset to fill the nan values.
-            Defaults to ``mean``.
+            Defaults to ``random``.
         model_missing_values (bool):
             **DEPRECATED** Whether to create a new column to indicate which values were null or
             not. The column will be created only if there are null values. If ``True``, create
@@ -53,14 +53,14 @@ class UnixTimestampEncoder(BaseTransformer):
 
     def __init__(
         self,
-        missing_value_replacement='mean',
+        missing_value_replacement='random',
         model_missing_values=None,
         datetime_format=None,
         missing_value_generation='random',
         enforce_min_max_values=False,
     ):
         super().__init__()
-        self._set_missing_value_replacement('mean', missing_value_replacement)
+        self._set_missing_value_replacement('random', missing_value_replacement)
         self._set_missing_value_generation(missing_value_generation)
         self.enforce_min_max_values = enforce_min_max_values
         if model_missing_values is not None:
@@ -248,7 +248,7 @@ class OptimizedTimestampEncoder(UnixTimestampEncoder):
             Indicate what to replace the null values with. If the strings ``'mean'`` or ``'mode'``
             are given, replace them with the corresponding aggregation, if ``'random'``, use
             random values from the dataset to fill the nan values.
-            Defaults to ``mean``.
+            Defaults to ``random``.
         model_missing_values (bool):
             **DEPRECATED** Whether to create a new column to indicate which values were null or
             not. The column will be created only if there are null values. If ``True``, create
@@ -275,7 +275,7 @@ class OptimizedTimestampEncoder(UnixTimestampEncoder):
 
     def __init__(
         self,
-        missing_value_replacement=None,
+        missing_value_replacement='random',
         model_missing_values=None,
         datetime_format=None,
         missing_value_generation='random',

tests/unit/transformers/test_datetime.py

@@ -45,6 +45,14 @@ def test___init__with_model_missing_values(self):
         assert transformer.missing_value_generation == 'random'
         assert transformer.datetime_format == '%M-%d-%Y'
 
+    def test_default_missing_value_replacement(self):
+        """Test the default value of missing_value_replacement is 'random'"""
+        # Run
+        transformer = UnixTimestampEncoder()
+
+        # Assert
+        assert transformer.missing_value_replacement == 'random'
+
     def test__convert_to_datetime(self):
         """Test the ``_convert_to_datetime`` method.
 
@@ -270,7 +278,7 @@ def test__fit(self, null_transformer_mock):
         """
         # Setup
         data = pd.to_datetime(['2020-01-01', '2020-02-01', '2020-03-01'])
-        transformer = UnixTimestampEncoder()
+        transformer = UnixTimestampEncoder(missing_value_replacement='mean')
 
         # Run
         transformer._fit(data)
@@ -604,6 +612,14 @@ def test___init__(self):
         assert transformer.divider is None
         assert transformer.null_transformer is None
 
+    def test_default_missing_value_replacement(self):
+        """Test the default value of missing_value_replacement is 'random'"""
+        # Run
+        transformer = OptimizedTimestampEncoder()
+
+        # Assert
+        assert transformer.missing_value_replacement == 'random'
+
     def test__find_divider(self):
         """Test the ``_find_divider`` method.