PERF: Series.transform speedups (GH6496)

doc/source/v0.14.1.txt

@@ -137,7 +137,7 @@ Performance
 
 
 - Improvements in dtype inference for numeric operations involving yielding performance gains for dtypes: ``int64``, ``timedelta64``, ``datetime64`` (:issue:`7223`)
-
+- Improvements in Series.transform for signifcant performance gains (:issue`6496`)
 
 
 

pandas/core/groupby.py

@@ -14,7 +14,7 @@
 from pandas.core.categorical import Categorical
 from pandas.core.frame import DataFrame
 from pandas.core.generic import NDFrame
-from pandas.core.index import Index, MultiIndex, _ensure_index
+from pandas.core.index import Index, MultiIndex, _ensure_index, _union_indexes
 from pandas.core.internals import BlockManager, make_block
 from pandas.core.series import Series
 from pandas.core.panel import Panel
@@ -425,7 +425,7 @@ def convert(key, s):
                 return Timestamp(key).asm8
             return key
 
-        sample = list(self.indices)[0]
+        sample = next(iter(self.indices))
         if isinstance(sample, tuple):
             if not isinstance(name, tuple):
                 raise ValueError("must supply a tuple to get_group with multiple grouping keys")
@@ -2193,33 +2193,37 @@ def transform(self, func, *args, **kwargs):
         -------
         transformed : Series
         """
-        result = self._selected_obj.copy()
-        if hasattr(result, 'values'):
-            result = result.values
-        dtype = result.dtype
+        dtype = self._selected_obj.dtype
 
         if isinstance(func, compat.string_types):
             wrapper = lambda x: getattr(x, func)(*args, **kwargs)
         else:
             wrapper = lambda x: func(x, *args, **kwargs)
 
-        for name, group in self:
+        result = self._selected_obj.values.copy()
+        for i, (name, group) in enumerate(self):
 
             object.__setattr__(group, 'name', name)
             res = wrapper(group)
+
             if hasattr(res, 'values'):
                 res = res.values
 
-            # need to do a safe put here, as the dtype may be different
-            # this needs to be an ndarray
-            result = Series(result)
-            result.iloc[self._get_index(name)] = res
-            result = result.values
+            # may need to astype
+            try:
+               common_type = np.common_type(np.array(res), result)
+               if common_type != result.dtype:
+                   result = result.astype(common_type)
+            except:
+                pass
+
+            indexer = self._get_index(name)
+            result[indexer] = res
 
-        # downcast if we can (and need)
         result = _possibly_downcast_to_dtype(result, dtype)
-        return self._selected_obj.__class__(result, index=self._selected_obj.index,
-                                  name=self._selected_obj.name)
+        return self._selected_obj.__class__(result,
+                                            index=self._selected_obj.index,
+                                            name=self._selected_obj.name)
 
     def filter(self, func, dropna=True, *args, **kwargs):
         """

pandas/tests/test_groupby.py

@@ -126,8 +126,10 @@ def checkit(dtype):
             assert_series_equal(agged, grouped.mean())
             assert_series_equal(grouped.agg(np.sum), grouped.sum())
 
+            expected = grouped.apply(lambda x: x * x.sum())
             transformed = grouped.transform(lambda x: x * x.sum())
             self.assertEqual(transformed[7], 12)
+            assert_series_equal(transformed, expected)
 
             value_grouped = data.groupby(data)
             assert_series_equal(value_grouped.aggregate(np.mean), agged)

vb_suite/groupby.py

@@ -376,3 +376,21 @@ def f(g):
 """
 
 groupby_transform = Benchmark("data.groupby(level='security_id').transform(f_fillna)", setup)
+
+setup = common_setup + """
+np.random.seed(0)
+
+N = 120000
+N_TRANSITIONS = 1400
+
+# generate groups
+transition_points = np.random.permutation(np.arange(N))[:N_TRANSITIONS]
+transition_points.sort()
+transitions = np.zeros((N,), dtype=np.bool)
+transitions[transition_points] = True
+g = transitions.cumsum()
+
+df = DataFrame({ 'signal' : np.random.rand(N)})
+"""
+
+groupby_transform2 = Benchmark("df['signal'].groupby(g).transform(np.mean)", setup)