嘗試將值分配給groupby對象的新列時的NotImplementedError

Question

減法工作並返回Series，其索引與groupby對象（一年中的月份，即1-12）相同。創建新列並將值分配給新列似乎會導致NotImplementedError。

我想從原來的dataframe減去相應月份的12個月值，即從1月份的每個數據點中減去1（1月）的值，等等。

test = df 
grouped = test.groupby(test.index.month) 
values_to_subtract = grouped['A'].median() - test['A'].median() 
print values_to_subtract 
grouped['new col'] = grouped['B'] - values_to_subtract 
print grouped['new col'] 

--------------------------------------------------------------------------- 
NotImplementedError      Traceback (most recent call last) 
<ipython-input-226-9bff427dc855> in <module>() 
     3 values_to_subtract = grouped['A'].median() - test['A'].median() 
     4 print values_to_subtract 
----> 5 grouped['new col'] = grouped['B'] - values_to_subtract 
     6 print grouped['new col'] 

C:\Users\AppData\Local\Enthought\Canopy32\User\lib\site-packages\pandas\core\ops.pyc in wrapper(left, right, name) 
    503    if hasattr(lvalues, 'values'): 
    504     lvalues = lvalues.values 
--> 505    return left._constructor(wrap_results(na_op(lvalues, rvalues)), 
    506          index=left.index, name=left.name, 
    507          dtype=dtype) 

C:\Users\AppData\Local\Enthought\Canopy32\User\lib\site-packages\pandas\core\ops.pyc in na_op(x, y) 
    443   try: 
    444    result = expressions.evaluate(op, str_rep, x, y, 
--> 445           raise_on_error=True, **eval_kwargs) 
    446   except TypeError: 
    447    if isinstance(y, (pa.Array, pd.Series)): 

C:\Users\AppData\Local\Enthought\Canopy32\User\lib\site-packages\pandas\computation\expressions.pyc in evaluate(op, op_str, a, b, raise_on_error, use_numexpr, **eval_kwargs) 
    210  if use_numexpr: 
    211   return _evaluate(op, op_str, a, b, raise_on_error=raise_on_error, 
--> 212       **eval_kwargs) 
    213  return _evaluate_standard(op, op_str, a, b, raise_on_error=raise_on_error) 
    214 

C:\Users\AppData\Local\Enthought\Canopy32\User\lib\site-packages\pandas\computation\expressions.pyc in _evaluate_standard(op, op_str, a, b, raise_on_error, **eval_kwargs) 
    63  if _TEST_MODE: 
    64   _store_test_result(False) 
---> 65  return op(a, b) 
    66 
    67 

C:\Users\AppData\Local\Enthought\Canopy32\User\lib\site-packages\pandas\core\ops.pyc in <lambda>(x, y) 
    70   rmul=arith_method(operator.mul, names('rmul'), op('*'), 
    71       default_axis=default_axis, reversed=True), 
---> 72   rsub=arith_method(lambda x, y: y - x, names('rsub'), op('-'), 
    73       default_axis=default_axis, reversed=True), 
    74   rtruediv=arith_method(lambda x, y: operator.truediv(y, x), 

C:\Users\AppData\Local\Enthought\Canopy32\User\lib\site-packages\pandas\core\groupby.pyc in __getitem__(self, key) 
    487 
    488  def __getitem__(self, key): 
--> 489   raise NotImplementedError 
    490 
    491  def _make_wrapper(self, name): 

NotImplementedError: 

1 -3.40 
2 -3.60 
3 -5.30 
4  0.15 
5  1.80 
6 -0.80 
7  2.15 
8  6.70 
9  3.90 
10 1.45 
11 -0.75 
12 -2.70 
Name: A, dtype: float64 

Answer 1

我想你想被使用transform這裏：

test['A'] - grouped['A'].transform("median") 

---

這是一些奇怪的代碼...

grouped = test.groupby(test.index.month) 
values_to_subtract = grouped['A'].median() - test['A'].median() 

現在，values_to_subtract是一個系列（假設只有一個'A'列，而grouped['B']是一個SeriesGrouby對象...它不會使感覺減去它們！

grouped['B'] - values_to_subtract 

而且，你不能分配給DataFrameGroupby對象列，所以即使上面是一個系列：

grouped['new col'] = _ 
TypeError: 'DataFrameGroupBy' object does not support item assignment