BUG: DataFrameGroupBy.transform and ngroup do not work with cumcount

doc/source/whatsnew/v0.25.1.rst

@@ -121,7 +121,7 @@ Groupby/resample/rolling
 - Bug in :meth:`pandas.core.groupby.DataFrameGroupBy.transform` where applying a timezone conversion lambda function would drop timezone information (:issue:`27496`)
 - Bug in windowing over read-only arrays (:issue:`27766`)
 - Fixed segfault in `pandas.core.groupby.DataFrameGroupBy.quantile` when an invalid quantile was passed (:issue:`27470`)
--
+- Bug in :meth:`pandas.core.groupby.DataFrameGroupBy.transform` where ``cumcount`` and ``ngroup`` fail (:issue:`27472` and :issue:`27468`)
 
 Reshaping
 ^^^^^^^^^

pandas/core/groupby/base.py

@@ -100,7 +100,9 @@ def _gotitem(self, key, ndim, subset=None):
 
 # cythonized transformations or canned "agg+broadcast", which do not
 # require postprocessing of the result by transform.
-cythonized_kernels = frozenset(["cumprod", "cumsum", "shift", "cummin", "cummax"])
+cythonized_kernels = frozenset(
+    ["cumprod", "cumsum", "shift", "cummin", "cummax", "cumcount"]
+)
 
 cython_cast_blacklist = frozenset(["rank", "count", "size", "idxmin", "idxmax"])
 
@@ -120,7 +122,6 @@ def _gotitem(self, key, ndim, subset=None):
         "mean",
         "median",
         "min",
-        "ngroup",
         "nth",
         "nunique",
         "prod",
@@ -158,6 +159,7 @@ def _gotitem(self, key, ndim, subset=None):
         "rank",
         "shift",
         "tshift",
+        "ngroup"
     ]
 )
 

pandas/core/groupby/generic.py

@@ -583,7 +583,8 @@ def transform(self, func, *args, **kwargs):
             if not (func in base.transform_kernel_whitelist):
                 msg = "'{func}' is not a valid function name for transform(name)"
                 raise ValueError(msg.format(func=func))
-            if func in base.cythonized_kernels:
+            # transformation are added as well since they are broadcasted already
+            if func in base.cythonized_kernels or func in base.transformation_kernels:
                 # cythonized transformation or canned "reduction+broadcast"
                 return getattr(self, func)(*args, **kwargs)
             else:

pandas/tests/groupby/test_transform.py

@@ -1034,8 +1034,6 @@ def test_transform_agg_by_name(reduction_func, obj):
     func = reduction_func
     g = obj.groupby(np.repeat([0, 1], 3))
 
-    if func == "ngroup":  # GH#27468
-        pytest.xfail("TODO: g.transform('ngroup') doesn't work")
     if func == "size":  # GH#27469
         pytest.xfail("TODO: g.transform('size') doesn't work")
 
@@ -1074,3 +1072,48 @@ def test_transform_lambda_with_datetimetz():
         name="time",
     )
     assert_series_equal(result, expected)
+
+
+def test_transform_cumcount_ngroup():
+    df = DataFrame(dict(a=[0, 0, 0, 1, 1, 1], b=range(6)))
+    g = df.groupby(np.repeat([0, 1], 3))
+
+    # GH 27472
+    result = g.transform("cumcount")
+    expected = g.cumcount()
+    assert_series_equal(result, expected)
+
+    # GH 27468
+    result = g.transform("ngroup")
+    expected = g.ngroup()
+    assert_series_equal(result, expected)
+
+
+@pytest.mark.parametrize(
+    "func",
+    [
+        "backfill",
+        "bfill",
+        "cumcount",
+        "cummax",
+        "cummin",
+        "cumprod",
+        "cumsum",
+        "diff",
+        "ffill",
+        "pad",
+        "pct_change",
+        "rank",
+        "shift",
+        "ngroup",
+    ],
+)
+def test_transformation_kernels_length(func):
+    # This test is to evaluate if after transformation, the index
+    # of transformed data is still the same with original DataFrame
+    # TODO: exceptions are fillna, tshfit and corrwith
+    df = DataFrame(dict(a=[0, 0, 0, 1, 1, 1], b=range(6)))
+    g = df.groupby(np.repeat([0, 1], 3))
+
+    result = g.transform(func)
+    assert (result.index == df.index).all()