BUG: fixes #53935 Categorical order lost after call to remove_categories (#54027)

* Changed the default value for sort to 'False' in the difference method. This allows the difference method to then call the _difference method and finally call the _maybe_try_sort method.  In the _maybe_try_sort_method it will sort the values if sort is not False.  That's why in the original code haveing sort=None would still sort the categories.  This way the code will only sort if you set sort=True.

* Added test to show the variable value change behaves the way we want it to.

* Added bug fix to whatsnew.

* Changed bug fix implementation to simply check if the Ordered value is set to True, if so it sets sort=False in the call to difference in remove_categories.

* Changed bug fix implementation to simply check if the Ordered value is set to True, if so it sets sort=False in the call to difference in remove_categories.

* Switched the implementation to a ternary to check for ordered. This seems to work better since we are not overriding a default argument this way.

Author: paulreece

doc/source/whatsnew/v2.1.0.rst

@@ -376,6 +376,7 @@ Bug fixes
 
 Categorical
 ^^^^^^^^^^^
+- Bug in :meth:`CategoricalIndex.remove_categories` where ordered categories would not be maintained (:issue:`53935`).
 - Bug in :meth:`Series.astype` with ``dtype="category"`` for nullable arrays with read-only null value masks (:issue:`53658`)
 - Bug in :meth:`Series.map` , where the value of the ``na_action`` parameter was not used if the series held a :class:`Categorical` (:issue:`22527`).
 -

pandas/core/arrays/categorical.py

@@ -1369,7 +1369,11 @@ def remove_categories(self, removals) -> Self:
             removals = [removals]
 
         removals = Index(removals).unique().dropna()
-        new_categories = self.dtype.categories.difference(removals)
+        new_categories = (
+            self.dtype.categories.difference(removals, sort=False)
+            if self.dtype.ordered is True
+            else self.dtype.categories.difference(removals)
+        )
         not_included = removals.difference(self.dtype.categories)
 
         if len(not_included) != 0:

pandas/tests/indexes/categorical/test_category.py

@@ -373,3 +373,18 @@ def test_method_delegation(self):
         msg = "cannot use inplace with CategoricalIndex"
         with pytest.raises(ValueError, match=msg):
             ci.set_categories(list("cab"), inplace=True)
+
+    def test_remove_maintains_order(self):
+        ci = CategoricalIndex(list("abcdda"), categories=list("abcd"))
+        result = ci.reorder_categories(["d", "c", "b", "a"], ordered=True)
+        tm.assert_index_equal(
+            result,
+            CategoricalIndex(list("abcdda"), categories=list("dcba"), ordered=True),
+        )
+        result = result.remove_categories(["c"])
+        tm.assert_index_equal(
+            result,
+            CategoricalIndex(
+                ["a", "b", np.nan, "d", "d", "a"], categories=list("dba"), ordered=True
+            ),
+        )