BUG: workaround for np.bincount with minlength=0

Author: behzadnouri

doc/source/whatsnew/v0.18.0.txt

@@ -106,6 +106,7 @@ Performance Improvements
 Bug Fixes
 ~~~~~~~~~
 
+- Bug in ``GroupBy.size`` when data-frame is empty. (:issue:`11699`)
 
 
 - Bug in ``.loc`` against ``CategoricalIndex`` may result in normal ``Index`` (:issue:`11586`)

pandas/core/categorical.py

@@ -1095,7 +1095,8 @@ def value_counts(self, dropna=True):
         ix, clean = np.arange(ncat), mask.all()
 
         if dropna or clean:
-            count = bincount(code if clean else code[mask], minlength=ncat)
+            obs = code if clean else code[mask]
+            count = bincount(obs, minlength=ncat or None)
         else:
             count = bincount(np.where(mask, code, ncat))
             ix = np.append(ix, -1)

pandas/core/groupby.py

@@ -1439,7 +1439,7 @@ def size(self):
         """
         ids, _, ngroup = self.group_info
         ids = com._ensure_platform_int(ids)
-        out = np.bincount(ids[ids != -1], minlength=ngroup)
+        out = np.bincount(ids[ids != -1], minlength=ngroup or None)
         return Series(out, index=self.result_index, dtype='int64')
 
     @cache_readonly
@@ -2822,7 +2822,7 @@ def count(self):
 
         mask = (ids != -1) & ~isnull(val)
         ids = com._ensure_platform_int(ids)
-        out = np.bincount(ids[mask], minlength=ngroups) if ngroups != 0 else []
+        out = np.bincount(ids[mask], minlength=ngroups or None)
 
         return Series(out, index=self.grouper.result_index, name=self.name, dtype='int64')
 

pandas/core/series.py

@@ -1142,7 +1142,8 @@ def count(self, level=None):
             lab[mask] = cnt = len(lev)
             lev = lev.insert(cnt, _get_na_value(lev.dtype.type))
 
-        out = np.bincount(lab[notnull(self.values)], minlength=len(lev))
+        obs = lab[notnull(self.values)]
+        out = np.bincount(obs, minlength=len(lev) or None)
         return self._constructor(out, index=lev, dtype='int64').__finalize__(self)
 
     def mode(self):

pandas/tests/test_groupby.py

@@ -2525,6 +2525,11 @@ def test_size(self):
             right = df.groupby(key, sort=sort)['c'].apply(lambda a: a.shape[0])
             assert_series_equal(left, right, check_names=False)
 
+        # GH11699
+        df = DataFrame([], columns=['A', 'B'])
+        out = Series([], dtype='int64', index=Index([], name='A'))
+        assert_series_equal(df.groupby('A').size(), out)
+
     def test_count(self):
         from string import ascii_lowercase
         n = 1 << 15