python - Pandas 属性错误: 'unicode' object has no attribute 'view'

Question

这是一个 killer 级问题，对于像我这样的 pandas 新手来说可能有一个简单的解决方案:

我正在尝试用该标签的最新版本(在单独的 DataFrame (latest_version) 中找到)替换 pandas DataFrame (df) 的一条记录。

df.ix[label] = latest_version.ix[label]

错误:

AttributeError: 'unicode' object has no attribute 'view'

df 本身又大又复杂(并且是专有的)，所以我想尽可能避免发布它；我希望有一些简单的东西我错过了，但我无法弄清楚。

编辑:df.info()和latest_version.info()的输出

ipdb> df.info()
<class 'pandas.core.frame.DataFrame'>
Index: 7 entries, A to G
Data columns (total 73 columns):
Column 0   7 non-null object
Column 1   7 non-null object
Column 2   7 non-null object
Column 3   7 non-null object
Column 4   7 non-null object
Column 5   7 non-null float64
Column 6   1 non-null object
Column 7   7 non-null object
Column 8   7 non-null object
Column 9   6 non-null datetime64[ns]
Column 10  0 non-null object
Column 11  0 non-null object
Column 12  5 non-null object
Column 13  0 non-null object
Column 14  0 non-null object
Column 15  6 non-null datetime64[ns]
Column 16  0 non-null object
Column 17  0 non-null object
Column 18  0 non-null object
Column 19  0 non-null object
Column 20  0 non-null object
Column 21  0 non-null object
Column 22  0 non-null object
Column 23  0 non-null object
Column 24  0 non-null object
Column 25  0 non-null object
Column 26  0 non-null object
Column 27  0 non-null object
Column 28  0 non-null object
Column 29  0 non-null object
Column 30  0 non-null object
Column 31  0 non-null object
Column 32  0 non-null object
Column 33  0 non-null object
Column 34  0 non-null object
Column 35  0 non-null object
Column 36  0 non-null object
Column 37  4 non-null object
Column 38  6 non-null object
Column 39  4 non-null object
Column 40  0 non-null object
Column 41  0 non-null object
Column 42  0 non-null object
Column 43  6 non-null object
Column 44  0 non-null object
Column 45  6 non-null object
Column 46  0 non-null object
Column 47  4 non-null object
Column 48  0 non-null object
Column 49  4 non-null object
Column 50  0 non-null object
Column 51  0 non-null object
Column 52  0 non-null object
Column 53  0 non-null object
Column 54  0 non-null object
Column 55  0 non-null object
Column 56  0 non-null object
Column 57  0 non-null object
Column 58  0 non-null object
Column 59  0 non-null object
Column 60  0 non-null object
Column 61  0 non-null object
Column 62  0 non-null object
Column 63  0 non-null object
Column 64  0 non-null object
Column 65  0 non-null object
Column 66  0 non-null object
Column 67  0 non-null object
Column 68  0 non-null object
Column 69  0 non-null object
Column 70  0 non-null object
Column 71  0 non-null object
Column 72  0 non-null object
dtypes: datetime64[ns](2), float64(1), object(70)ipdb> 

ipdb> latest_version.info()
<class 'pandas.core.frame.DataFrame'>
Index: 4 entries, A to D
Data columns (total 73 columns):
Column 0   4 non-null object
Column 1   4 non-null object
Column 2   4 non-null object
Column 3   4 non-null object
Column 4   4 non-null object
Column 5   4 non-null int64
Column 6   4 non-null object
Column 7   4 non-null object
Column 8   4 non-null object
Column 9   4 non-null object
Column 10  4 non-null object
Column 11  4 non-null object
Column 12  4 non-null object
Column 13  4 non-null object
Column 14  4 non-null object
Column 15  4 non-null object
Column 16  3 non-null object
Column 17  4 non-null object
Column 18  4 non-null object
Column 19  4 non-null object
Column 20  3 non-null object
Column 21  3 non-null object
Column 22  4 non-null object
Column 23  4 non-null object
Column 24  4 non-null object
Column 25  4 non-null object
Column 26  4 non-null object
Column 27  4 non-null object
Column 28  4 non-null object
Column 29  4 non-null object
Column 30  4 non-null object
Column 31  4 non-null object
Column 32  4 non-null object
Column 33  4 non-null object
Column 34  4 non-null object
Column 35  4 non-null object
Column 36  4 non-null object
Column 37  4 non-null object
Column 38  4 non-null object
Column 39  4 non-null object
Column 40  4 non-null object
Column 41  4 non-null object
Column 42  4 non-null object
Column 43  4 non-null object
Column 44  4 non-null object
Column 45  4 non-null float64
Column 46  4 non-null object
Column 47  4 non-null object
Column 48  4 non-null object
Column 49  4 non-null object
Column 50  4 non-null object
Column 51  4 non-null object
Column 52  4 non-null object
Column 53  4 non-null object
Column 54  4 non-null object
Column 55  4 non-null object
Column 56  1 non-null object
Column 57  1 non-null object
Column 58  4 non-null object
Column 59  4 non-null object
Column 60  4 non-null object
Column 61  4 non-null object
Column 62  4 non-null object
Column 63  4 non-null object
Column 64  4 non-null object
Column 65  4 non-null object
Column 66  4 non-null object
Column 67  4 non-null object
Column 68  4 non-null object
Column 69  4 non-null object
Column 70  4 non-null object
Column 71  4 non-null object
Column 72  4 non-null object
dtypes: float64(1), int64(1), object(71)ipdb> 

进一步编辑(回应 Ed):以下是仅包含不同类型列的表格:

ipdb> latest_version.ix[:,[5,9,15]]
                                    line_number  entry_date entry_ref_a
unique_index                                                           
NEW/AAAAAAAAAAAAAAAAAAA                       0  2014-12-30  2015-01-14
NEW/AAAAAAAAAAAAAAAAAAB                       1  2014-12-30            
NEW/AAAAAAAAAAAAAAAAAAC                       2  2014-12-30            
ipdb>/df.ix[:,[5,9,15]]
                                    line_number           entry_date  \
unique_index                                                           
OLD/204442                                    0  1419897600000000000   
OLD/343278                                    1  1419897600000000000   
OLD/359628                                    2  1419897600000000000   
NEW/AAAAAAAAAAAAAAAAAAA                       0           2014-12-30   

                                            entry_ref_a  
unique_index                                             
OLD/204442                          1421193600000000000  
OLD/343278                          1421193600000000000  
OLD/359628                          1422230400000000000  
NEW/AAAAAAAAAAAAAAAAAAA                      2015-01-14  

绝对证实了这里存在类型不匹配问题的想法......

Answer 1

所以你的问题似乎是你试图分配的两个 dfs 之间的 dtypes 不匹配:

df dtypes: datetime64[ns](2), float64(1), object(70) 

同时

latest_version is :dtypes: float64(1), int64(1), object(71)

从输出中我们可以看到，与某些冲突的列是日期时间，而在另一个 df 的相应列中它们是 int64。

您可以通过执行以下操作将格式错误的列转换为日期时间:

df['entry_date'] = pd.to_datetime(df['entry_date')

同样适用于entry_ref_a