Small fixes for computing spreadfn aggregates

holoviews/core/data/__init__.py

@@ -488,7 +488,7 @@ def aggregate(self, dimensions=None, function=None, spreadfn=None, **kwargs):
             combined = self.clone(aggregated, kdims=kdims, new_type=Dataset)
             for i, d in enumerate(vdims):
                 dim = d('_'.join([d.name, spread_name]))
-                dvals = error.dimension_values(d, False, False)
+                dvals = error.dimension_values(d, flat=False)
                 combined = combined.add_dimension(dim, ndims+i, dvals, True)
             return combined.clone(new_type=Dataset if generic_type else type(self))
 

holoviews/core/data/pandas.py

@@ -139,11 +139,17 @@ def aggregate(cls, columns, dimensions, function, **kwargs):
         cols = [d.name for d in columns.kdims if d in dimensions]
         vdims = columns.dimensions('value', label='name')
         reindexed = data[cols+vdims]
+        if function in [np.std, np.var]:
+            # Fix for consistency with other backend
+            # pandas uses ddof=1 for std and var
+            fn = lambda x: function(x, ddof=0)
+        else:
+            fn = function
         if len(dimensions):
             grouped = reindexed.groupby(cols, sort=False)
-            return grouped.aggregate(function, **kwargs).reset_index()
+            return grouped.aggregate(fn, **kwargs).reset_index()
         else:
-            agg = reindexed.apply(function, **kwargs)
+            agg = reindexed.apply(fn, **kwargs)
             return pd.DataFrame.from_items([(col, [v]) for col, v in
                                             zip(agg.index, agg.values)])
 

tests/testdataset.py

@@ -397,6 +397,15 @@ def test_dataset_2D_partial_reduce_ht(self):
                           kdims=['x'], vdims=['z'])
         self.assertEqual(dataset.reduce(['y'], np.mean), reduced)
 
+    def test_dataset_2D_aggregate_spread_fn_with_duplicates(self):
+        dataset = Dataset({'x': np.array([0, 0, 1, 1]), 'y': np.array([0, 1, 2, 3]),
+                           'z': np.array([1, 2, 3, 4])},
+                          kdims=['x', 'y'], vdims=['z'])
+        agg = dataset.aggregate('x', function=np.mean, spreadfn=np.var)
+        self.assertEqual(agg, Dataset({'x': np.array([0, 1]), 'z': np.array([1.5, 3.5]),
+                                       'z_var': np.array([0.25, 0.25])},
+                                      kdims=['x'], vdims=['z', 'z_var']))
+
     def test_dataset_aggregate_ht(self):
         aggregated = Dataset({'Gender':['M', 'F'], 'Weight':[16.5, 10], 'Height':[0.7, 0.8]},
                              kdims=self.kdims[:1], vdims=self.vdims)
@@ -584,6 +593,9 @@ def test_dataset_add_dimensions_values_hm(self):
     def test_dataset_add_dimensions_values_ht(self):
         raise SkipTest("Not supported")
 
+    def test_dataset_2D_aggregate_spread_fn_with_duplicates(self):
+        raise SkipTest("Not supported")
+
     def test_dataset_sort_hm(self):
         raise SkipTest("Not supported")