PERF: Correct signature for group_nth / group_object (pandas-dev#19579)

asv_bench/benchmarks/groupby.py

@@ -160,6 +160,22 @@ def time_series_nth(self, df):
         df[1].groupby(df[0]).nth(0)
 
 
+class NthObject(object):
+
+    goal_time = 0.2
+
+    def setup_cache(self):
+        df = DataFrame(np.random.randint(1, 100, (10000,)), columns=['g'])
+        df['obj'] = ['a'] * 5000 + ['b'] * 5000
+        return df
+
+    def time_nth(self, df):
+        df.groupby('g').nth(5)
+
+    def time_nth_last(self, df):
+        df.groupby('g').last()
+
+
 class DateAttributes(object):
 
     goal_time = 0.2

doc/source/whatsnew/v0.23.0.txt

@@ -746,6 +746,7 @@ Groupby/Resample/Rolling
 - Bug in :func:`DataFrame.transform` where particular aggregation functions were being incorrectly cast to match the dtype(s) of the grouped data (:issue:`19200`)
 - Bug in :func:`DataFrame.groupby` passing the `on=` kwarg, and subsequently using ``.apply()`` (:issue:`17813`)
 - Bug in :func:`DataFrame.resample().aggregate` not raising a ``KeyError`` when aggregating a non-existent column (:issue:`16766`, :issue:`19566`)
+- Fixed a performance regression for ``GroupBy.nth`` and ``GroupBy.last`` with some object columns (:issue:`19283`)
 
 Sparse
 ^^^^^^

pandas/_libs/groupby.pyx

@@ -36,7 +36,8 @@ def group_nth_object(ndarray[object, ndim=2] out,
                      ndarray[int64_t] counts,
                      ndarray[object, ndim=2] values,
                      ndarray[int64_t] labels,
-                     int64_t rank):
+                     int64_t rank,
+                     Py_ssize_t min_count=-1):
     """
     Only aggregates on axis=0
     """
@@ -47,6 +48,8 @@ def group_nth_object(ndarray[object, ndim=2] out,
         ndarray[int64_t, ndim=2] nobs
         ndarray[object, ndim=2] resx
 
+    assert min_count == -1, "'min_count' only used in add and prod"
+
     nobs = np.zeros((<object> out).shape, dtype=np.int64)
     resx = np.empty((<object> out).shape, dtype=object)
 
@@ -80,7 +83,8 @@ def group_nth_object(ndarray[object, ndim=2] out,
 def group_last_object(ndarray[object, ndim=2] out,
                       ndarray[int64_t] counts,
                       ndarray[object, ndim=2] values,
-                      ndarray[int64_t] labels):
+                      ndarray[int64_t] labels,
+                      Py_ssize_t min_count=-1):
     """
     Only aggregates on axis=0
     """
@@ -91,6 +95,8 @@ def group_last_object(ndarray[object, ndim=2] out,
         ndarray[object, ndim=2] resx
         ndarray[int64_t, ndim=2] nobs
 
+    assert min_count == -1, "'min_count' only used in add and prod"
+
     nobs = np.zeros((<object> out).shape, dtype=np.int64)
     resx = np.empty((<object> out).shape, dtype=object)