BUG: in _nsorted for frame with duplicated values index

Author: Tux1

doc/source/whatsnew/v0.19.0.txt

@@ -81,3 +81,5 @@ Performance Improvements
 
 Bug Fixes
 ~~~~~~~~~
+
+- Bug in ``DataFrame.nlargest`` and ``DataFrame.nsmallest`` when data-frame has duplicated value index. (:issue:`13412`)

pandas/core/frame.py

@@ -3315,10 +3315,24 @@ def _nsorted(self, columns, n, method, keep):
         if not com.is_list_like(columns):
             columns = [columns]
         columns = list(columns)
-        ser = getattr(self[columns[0]], method)(n, keep=keep)
+
+        df = self
+        index_unique = df.index.is_unique
+
+        if not index_unique:
+            df = df.reset_index()
+
+        ser = getattr(df[columns[0]], method)(n, keep=keep)
         ascending = dict(nlargest=False, nsmallest=True)[method]
-        return self.loc[ser.index].sort_values(columns, ascending=ascending,
-                                               kind='mergesort')
+        df = df.loc[ser.index]
+
+        if not index_unique:
+            index_names = self.index.names
+            nl = self.index.nlevels
+            df = df.set_index(df.columns[:nl].tolist())
+            df.index = df.index.set_names(index_names)
+
+        return df.sort_values(columns, ascending=ascending, kind='mergesort')
 
     def nlargest(self, n, columns, keep='first'):
         """Get the rows of a DataFrame sorted by the `n` largest

pandas/tests/frame/test_analytics.py

@@ -1253,6 +1253,30 @@ def test_nsmallest_multiple_columns(self):
         expected = df.sort_values(['a', 'c']).head(5)
         tm.assert_frame_equal(result, expected)
 
+    def test_nsorted_duplicated_index(self):
+        # GH13412
+        df = pd.DataFrame({'a': [1, 2, 3, 4],
+                           'b': [4, 3, 2, 1]}, index=[0, 0, 1, 1])
+        result = df.nlargest(1, 'a')
+        result2 = df.nlargest(2, 'a')
+        expected = df.sort_values('a', ascending=False).head(1)
+        expected2 = df.sort_values('a', ascending=False).head(2)
+        tm.assert_frame_equal(result, expected)
+        tm.assert_frame_equal(result2, expected2)
+
+    def test_nsorted_multiple_index(self):
+        # GH13412
+        df = pd.DataFrame({'a': [1, 2, 3, 4],
+                           'b': [4, 3, 2, 1],
+                           'c': [4, 3, 2, 1]}, index=[0, 0, 1, 1])
+        df = df.reset_index().set_index(['index', 'c'])
+        result = df.nlargest(1, 'a')
+        result2 = df.nlargest(2, 'a')
+        expected = df.sort_values('a', ascending=False).head(1)
+        expected2 = df.sort_values('a', ascending=False).head(2)
+        tm.assert_frame_equal(result, expected)
+        tm.assert_frame_equal(result2, expected2)
+
     # ----------------------------------------------------------------------
     # Isin