Changed the target imputer to just look for all nulls in the target a…

…nd raise.

evalml/pipelines/components/transformers/imputers/target_imputer.py

@@ -74,15 +74,12 @@ def fit(self, X, y):
         Returns:
             self
         """
-        from woodwork.logical_types import Unknown
-
         if y is None:
             return self
         y = infer_feature_types(y)
-        if isinstance(y.ww.logical_type, Unknown):
-            raise TypeError("Provided target full of pd.NA.")
+        if all(y.isnull()):
+            raise TypeError("Provided target full of nulls.")
         y = y.to_frame()
-        # should y be an un-inited dataframe?
 
         # Convert all bool dtypes to category for fitting
         if (y.dtypes == bool).all():

evalml/tests/component_tests/test_target_imputer.py

@@ -134,11 +134,11 @@ def test_target_imputer_fit_transform_all_nan_empty(y):
 
     imputer = TargetImputer()
 
-    with pytest.raises(TypeError, match="Provided target full of pd.NA."):
+    with pytest.raises(TypeError, match="Provided target full of nulls."):
         imputer.fit(None, y)
 
     imputer = TargetImputer()
-    with pytest.raises(TypeError, match="Provided target full of pd.NA."):
+    with pytest.raises(TypeError, match="Provided target full of nulls."):
         imputer.fit_transform(None, y)
 
 

evalml/tests/pipeline_tests/test_pipeline_utils.py

@@ -203,7 +203,7 @@ def test_make_pipeline(
             )
             imputer = (
                 []
-                if (column_names == ["dates"] and input_type == "ww")
+                if (column_names in [["dates"], ["all_null"]] and input_type == "ww")
                 or ((column_names in [["text"], ["dates"]]) and input_type == "pd")
                 else [Imputer]
             )