From 5a4f9a61a4fdbc26b5a3bb233439bebe136627c4 Mon Sep 17 00:00:00 2001
From: Loc Nguyen <locnb185@gmail.com>
Date: Fri, 30 Oct 2026 08:46:23 +0700
Subject: [PATCH] Do not re-add an existing category in CategoricalImputer
 (#1113)

---
 feature_engine/imputation/categorical.py      |  8 ++++--
 .../test_categorical_imputer.py               | 27 +++++++++++++++++++
 2 files changed, 33 insertions(+), 2 deletions(-)

diff --git a/feature_engine/imputation/categorical.py b/feature_engine/imputation/categorical.py
index 2cd4a00a3..995400c57 100644
--- a/feature_engine/imputation/categorical.py
+++ b/feature_engine/imputation/categorical.py
@@ -243,9 +243,13 @@ def transform(self, X: pd.DataFrame) -> pd.DataFrame:
             X = self._transform(X)
 
             # if variable is of type category, we need to add the new
-            # category, before filling in the nan
+            # category, before filling in the nan, unless it is already
+            # one of the categories
             for variable in self.variables_:
-                if X[variable].dtype.name == "category":
+                if (
+                    X[variable].dtype.name == "category"
+                    and self.imputer_dict_[variable] not in X[variable].cat.categories
+                ):
                     X[variable] = X[variable].cat.add_categories(
                         self.imputer_dict_[variable]
                     )
diff --git a/tests/test_imputation/test_categorical_imputer.py b/tests/test_imputation/test_categorical_imputer.py
index 182e8826b..2f4e1bfbc 100644
--- a/tests/test_imputation/test_categorical_imputer.py
+++ b/tests/test_imputation/test_categorical_imputer.py
@@ -294,6 +294,33 @@ def test_variables_cast_as_category_frequent(df_na):
     pd.testing.assert_frame_equal(X_transformed, X_reference)
 
 
+@pytest.mark.parametrize("ordered", [False, True])
+def test_variables_cast_as_category_with_fill_value_already_a_category(ordered):
+    categories = ["unknown", "west", "east"]
+    X = pd.DataFrame(
+        {
+            "region": pd.Categorical(
+                ["east", None, "west"], categories=categories, ordered=ordered
+            )
+        }
+    )
+
+    imputer = CategoricalImputer(fill_value="unknown", variables=["region"])
+    X_transformed = imputer.fit_transform(X)
+
+    X_reference = pd.DataFrame(
+        {
+            "region": pd.Categorical(
+                ["east", "unknown", "west"], categories=categories, ordered=ordered
+            )
+        }
+    )
+    pd.testing.assert_frame_equal(X_transformed, X_reference)
+
+    # transforming the imputed data again leaves it unchanged
+    pd.testing.assert_frame_equal(imputer.transform(X_transformed), X_reference)
+
+
 @pytest.mark.parametrize(
     "ignore_format",
     [22.3, 1, "HOLA", {"key1": "value1", "key2": "value2", "key3": "value3"}],