From 8bae0d42db5e4fbd93407a016b35267f830a4ace Mon Sep 17 00:00:00 2001
From: dsm054 <dsm054@gmail.com>
Date: Sun, 22 Mar 2015 11:38:18 -0400
Subject: [PATCH] BUG: ensure we use group sizes, not group counts, in
 transform (GH9697)

---
 doc/source/whatsnew/v0.16.1.txt |  2 ++
 pandas/core/groupby.py          |  2 +-
 pandas/tests/test_groupby.py    | 13 +++++++++++++
 3 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/doc/source/whatsnew/v0.16.1.txt b/doc/source/whatsnew/v0.16.1.txt
index bf01d3b21f3fa..5801d1b811790 100644
--- a/doc/source/whatsnew/v0.16.1.txt
+++ b/doc/source/whatsnew/v0.16.1.txt
@@ -31,3 +31,5 @@ Performance Improvements
 
 Bug Fixes
 ~~~~~~~~~
+
+- Bug in ``transform`` causing length mismatch when null entries were present and a fast aggregator was being used (:issue:`9697`)
diff --git a/pandas/core/groupby.py b/pandas/core/groupby.py
index 73439fb1e535d..6d98b3b99021b 100644
--- a/pandas/core/groupby.py
+++ b/pandas/core/groupby.py
@@ -2453,7 +2453,7 @@ def _transform_fast(self, func):
         if isinstance(func, compat.string_types):
             func = getattr(self,func)
         values = func().values
-        counts = self.count().values
+        counts = self.size().values
         values = np.repeat(values, com._ensure_platform_int(counts))
 
         return self._set_result_index_ordered(Series(values))
diff --git a/pandas/tests/test_groupby.py b/pandas/tests/test_groupby.py
index 79ebb80fc9ebb..e7001eb09f20c 100644
--- a/pandas/tests/test_groupby.py
+++ b/pandas/tests/test_groupby.py
@@ -1058,6 +1058,19 @@ def test_transform_function_aliases(self):
         expected = self.df.groupby('A')['C'].transform(np.mean)
         assert_series_equal(result, expected)
 
+    def test_transform_length(self):
+        # GH 9697
+        df = pd.DataFrame({'col1':[1,1,2,2], 'col2':[1,2,3,np.nan]})
+        expected = pd.Series([3.0]*4)
+        def nsum(x):
+            return np.nansum(x)
+        results = [df.groupby('col1').transform(sum)['col2'],
+                   df.groupby('col1')['col2'].transform(sum),
+                   df.groupby('col1').transform(nsum)['col2'],
+                   df.groupby('col1')['col2'].transform(nsum)]
+        for result in results:
+            assert_series_equal(result, expected)
+
     def test_with_na(self):
         index = Index(np.arange(10))