python - 为什么字符串方法在 for 循环中使用时停止对对象列起作用

Question

我正在 jupyter 笔记本中执行此代码。

我正在尝试更改数据框中的对象类型，如下所示:

column 1
#Green #Blue #Orange
#Green #Red
#Blue #Orange
#Orange

对此:

column 1
[Green, Blue, Orange]
[Green, Red]
[Blue, Orange]
[Orange]

当我尝试使用以下代码在一个数据帧中的一列上使用字符串方法时，它可以工作。

df1['column 1'] = df1['column 1'].str.replace('#', ' ') 

df1['column 1'] = df1['column 1'].str.split(',')

但是当我尝试将此过程压缩为多个数据帧中同一列的循环时，我收到属性错误(即 AttributeError: Can only use .str accessor with string values, which use np.object_ dtype在 Pandas 中):

df_list = [df1, df2, df3, df4, df5]

for df in df_list:
    df['column 1'] = df['column 1'].str.replace('#', ' ') 
    df['column 1'] = df['column 1'].str.split(',')

当它们本质上是相同的过程时为什么会出现这种情况？

这是我的 jupyter 笔记本中的回溯错误:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-276-51c1ece881fa> in <module>
      4 for df in df_list:
      5     df['player_tags'] = df['player_tags'].str.replace('#', ' ')
----> 6     df['player_tags'] = df['player_tags'].str.split(',')

~/anaconda3/lib/python3.6/site-packages/pandas/core/generic.py in __getattr__(self, name)
   3608         if (name in self._internal_names_set or name in self._metadata or
   3609                 name in self._accessors):
-> 3610             return object.__getattribute__(self, name)
   3611         else:
   3612             if name in self._info_axis:

~/anaconda3/lib/python3.6/site-packages/pandas/core/accessor.py in __get__(self, instance, owner)
     52             # this ensures that Series.str.<method> is well defined
     53             return self.accessor_cls
---> 54         return self.construct_accessor(instance)
     55 
     56     def __set__(self, instance, value):

~/anaconda3/lib/python3.6/site-packages/pandas/core/strings.py in _make_accessor(cls, data)
   1908             # (instead of test for object dtype), but that isn't practical for
   1909             # performance reasons until we have a str dtype (GH 9343)
-> 1910             raise AttributeError("Can only use .str accessor with string "
   1911                                  "values, which use np.object_ dtype in "
   1912                                  "pandas")

AttributeError: Can only use .str accessor with string values, which use np.object_ dtype in pandas

Answer 1

根据您提供的代码 - 它应该可以工作(事实上，必须工作，假设您的所有项目都是合法的)。

您始终可以获取示例代码并在您端进行检查:

df = pd.DataFrame({"column 1":['#Green #Blue #Orange','#Green #Red','#Blue #Orange','#Orange']})

df1 = df.copy()
df2 = df.copy()
df3 = df.copy()

df_list = [df1, df2, df3]

for df in df_list:
    df['column 1'] = df['column 1'].str.replace('#', ' ') 
    df['column 1'] = df['column 1'].str.split(',')

效果非常好!结论 - 在您的代码 df_list 中，您发送了相同的 df 两次(或更多)。示例 df2 是对 df1 的引用，这会给您带来问题。请注意，在我上面的代码中，df1 df2 df3 被复制(与引用相反)

检查一下(这将导致错误):

df = pd.DataFrame({"column 1":['#Green #Blue #Orange','#Green #Red','#Blue #Orange','#Orange']})
df['column 1'] = df['column 1'].str.replace('#', ' ')
df['column 1'] = df['column 1'].str.split(',')

然后再做一次:

df['column 1'] = df['column 1'].str.replace('#', ' ')
df['column 1'] = df['column 1'].str.split(',')

您将收到错误消息。