Passed constructor arguments of smart transformers to _get_transforme…

…r() method by default.

foreshadow/tests/test_transformers/test_smart.py

@@ -183,3 +183,14 @@ def test_preprocessor_hashencoder_no_name_collision():
     # since the number of categories for each column are above 30, HashingEncoder will be used with 30 components. The transformed
     # output should have in total 60 unique names.
     assert len(set(output.columns)) == 60
+
+
+def test_smart_encoder_more_than_30_levels_with_overwritten_cutoff():
+    import numpy as np
+    from foreshadow.transformers.smart import Encoder
+    from foreshadow.transformers.internals import OneHotEncoder
+
+    np.random.seed(0)
+    gt_30_random_data = np.random.choice(31, size=500)
+    smart_coder = Encoder(unique_num_cutoff=35)
+    assert isinstance(smart_coder.fit(gt_30_random_data), OneHotEncoder)

foreshadow/transformers/base.py

@@ -397,7 +397,7 @@ def fit(self, X, y=None, **kwargs):
             self.transformer = _Empty()
             self._set_to_empty = True
         else:
-            self._verify_transformer(X, y, refit=True)
+            self._verify_transformer(X, y, refit=True, **self.kwargs)
         inject = _inject_df(self.transformer, kwargs.pop("full_df", None))
         return self.transformer.fit(X, y, **{**kwargs, **inject})