Question

我有一个pandas数据帧df：

>>> df
                   net_inventory      sales    yr temp_stk_id
STK_ID   RPT_Date                                            
34_STK43 20080331       282.1603   359.4644  2008    34_STK43
         20080630       297.4760   716.7633  2008    34_STK43
         20080930       283.6312  1105.6332  2008    34_STK43
         20081231       296.9090  1380.4886  2008    34_STK43
         20090331       276.2348   363.3449  2009    34_STK43
         20090630       288.0186   753.6347  2009    34_STK43
         20090930       287.0811  1173.3760  2009    34_STK43


>>> df.dtypes
net_inventory    float64
sales            float64
yr                object
temp_stk_id       object

然后我想为“net_inventory”和“sales”列做一个“expanding_mean（）”，groupby ['temp_stk_id','yr']：

>>> df.groupby(['temp_stk_id','yr']).apply(lambda x: pd.expanding_mean(x))

它给出了以下错误：

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "D:\Python\Lib\site-packages\pandas\core\groupby.py", line 322, in apply
    return self._python_apply_general(f)
  File "D:\Python\Lib\site-packages\pandas\core\groupby.py", line 325, in _python_apply_general
    keys, values, mutated = self.grouper.apply(f, self.obj, self.axis)
  File "D:\Python\Lib\site-packages\pandas\core\groupby.py", line 596, in apply
    res = f(group)
  File "D:\Python\Lib\site-packages\pandas\core\groupby.py", line 321, in <lambda>
    f = lambda g: func(g, *args, **kwargs)
  File "<stdin>", line 1, in <lambda>
  File "D:\Python\Lib\site-packages\pandas\stats\moments.py", line 697, in f
    time_rule=time_rule, **kwargs)
  File "D:\Python\Lib\site-packages\pandas\stats\moments.py", line 281, in _rolling_moment
    return_hook, values = _process_data_structure(arg)
  File "D:\Python\Lib\site-packages\pandas\stats\moments.py", line 326, in _process_data_structure
    values = values.astype(float)
ValueError: invalid literal for float(): 34_STK43

我认为这是因为expanding_mean（）不适用于对象类型temp_stk_id和'yr'列，所以我想：

df.groupby(['temp_stk_id','yr']).apply(lambda x: pd.expanding_mean(x) for all columns except [`temp_stk_id`, 'yr'])

或

df.groupby(['temp_stk_id','yr']).apply(lambda x: pd.expanding_mean(x) for specified columns)

怎么做？

Answer 1

这看起来像是transform的工作：

In [10]: g = df1.groupby(['temp_stk_id','yr'])

In [11]: g.transform(pd.expanding_mean)
Out[11]: 
                   net_inventory       sales
STK_ID   RPT_Date                           
34_STK43 20080331     282.160300  359.464400
         20080630     289.818150  538.113850
         20080930     287.755833  727.286967
         20081231     290.044125  890.587375
         20090331     276.234800  363.344900
         20090630     282.126700  558.489800
         20090930     283.778167  763.451867

如何做Pandas dataframe.groupby（these_columns）.apply（some_columns的功能）？

1 个答案: