BUG: Sereis.nlargest thinks True < False

- use 1-arr insteat of -arr for sorting a boolean array casted to uint
- add test
- add whatsnew

Author: swyoon

doc/source/whatsnew/v0.25.0.rst

@@ -419,7 +419,7 @@ Other
 
 - Improved :class:`Timestamp` type checking in various datetime functions to prevent exceptions when using a subclassed `datetime` (:issue:`25851`)
 - Bug in :class:`Series` and :class:`DataFrame` repr where ``np.datetime64('NaT')`` and ``np.timedelta64('NaT')`` with ``dtype=object`` would be represented as ``NaN`` (:issue:`25445`)
--
+- Bug in :meth:`Series.nlargest` treats ``True`` as smaller than ``False`` (:issue:`26154`)
 -
 
 

pandas/core/algorithms.py

@@ -1118,6 +1118,10 @@ def compute(self, method):
                 # GH 21426: ensure reverse ordering at boundaries
                 arr -= 1
 
+            if is_bool_dtype(pandas_dtype):
+                # GH 26154: ensure False is smaller than True
+                arr = 1 - (-arr)
+
         if self.keep == 'last':
             arr = arr[::-1]
 

pandas/tests/series/test_analytics.py

@@ -1330,6 +1330,13 @@ def test_duplicate_keep_all_ties(self):
         expected = Series([6, 7, 7, 7, 7], index=[7, 3, 4, 5, 6])
         assert_series_equal(result, expected)
 
+    def test_boolean(self):
+        # GH 26154 : ensure True > False
+        s = Series([True, False])
+        result = s.nlargest(1)
+        expected = Series([True])
+        assert_series_equal(result, expected)
+
 
 class TestCategoricalSeriesAnalytics: