如何将 json 字符串的 pandas 列转换为新的数据框答案

【问题标题】：How to convert a pandas column of json strings into a new dataframe如何将 json 字符串的 pandas 列转换为新的数据框
【发布时间】：2016-09-28 13:06:56
【问题描述】：

我收到了一个 csv，其中的一列数据是一个 json 字符串，加载到 pandas 中时如下所示：

df.registered_office.head()


3    {u"country": u"United Kingdom", u"region": None, u"postal_code": u"ST6 3HL", u"street_address": u"Martin Leake House Waterloo Road\nCobridge", u"locality": u"Stoke On Trent"}
5             {u"country": u"England", u"region": None, u"postal_code": u"BS32 4JY", u"street_address": u"Woodlands Grange Woodlands Lane\nBradley Stoke", u"locality": u"Bristol"}
7       {u"country": u"United Kingdom", u"region": u"Staffordshire", u"postal_code": u"ST6 8JD", u"street_address": u"136 Knypersley Road\nNorton", u"locality": u"Stoke On Trent"}
8    {u"country": u"United Kingdom", u"region": u"Staffordshire", u"postal_code": u"ST9 9HQ", u"street_address": u"Ashlands\n253 Leek Road, Endon", u"locality": u"Stoke On Trent"}
9                      {u"country": u"United Kingdom", u"region": None, u"postal_code": u"ST1 5TA", u"street_address": u"C/O Kpmg Festival Way\nStoke On Trent", u"locality": None}
Name: registered_address, dtype: object

根据各种键将此列转换为新数据框或新列的最 Pythonic 方式是什么？

【问题讨论】：

标签： json csv pandas

【解决方案1】：

我觉得你可以用DataFrame.from_records:

df = pd.DataFrame({'registered_office':[
{u"country": u"United Kingdom", u"region": None, u"postal_code": u"ST6 3HL", u"street_address": u"Martin Leake House Waterloo Road\nCobridge", u"locality": u"Stoke On Trent"},
{u"country": u"England", u"region": None, u"postal_code": u"BS32 4JY", u"street_address": u"Woodlands Grange Woodlands Lane\nBradley Stoke", u"locality": u"Bristol"},
{u"country": u"United Kingdom", u"region": u"Staffordshire", u"postal_code": u"ST6 8JD", u"street_address": u"136 Knypersley Road\nNorton", u"locality": u"Stoke On Trent"},
{u"country": u"United Kingdom", u"region": u"Staffordshire", u"postal_code": u"ST9 9HQ", u"street_address": u"Ashlands\n253 Leek Road, Endon", u"locality": u"Stoke On Trent"},
{u"country": u"United Kingdom", u"region": None, u"postal_code": u"ST1 5TA", u"street_address": u"C/O Kpmg Festival Way\nStoke On Trent", u"locality": None}
]})

print (pd.DataFrame.from_records(df.registered_office.values.tolist()))
          country        locality postal_code         region  \
0  United Kingdom  Stoke On Trent     ST6 3HL           None   
1         England         Bristol    BS32 4JY           None   
2  United Kingdom  Stoke On Trent     ST6 8JD  Staffordshire   
3  United Kingdom  Stoke On Trent     ST9 9HQ  Staffordshire   
4  United Kingdom            None     ST1 5TA           None   

                                   street_address  
0      Martin Leake House Waterloo Road\nCobridge  
1  Woodlands Grange Woodlands Lane\nBradley Stoke  
2                     136 Knypersley Road\nNorton  
3                  Ashlands\n253 Leek Road, Endon  
4           C/O Kpmg Festival Way\nStoke On Trent

【讨论】：

感谢您的快速回复。它几乎可以工作，但它将字符串中的单个字符分成列，而不是键。
所以你需要转置？ print (pd.DataFrame.from_records(df.registered_office.values.tolist()).T) ?
你的输出不一样？
对不起。我失去了网络连接。这是我的回复：对不起，我可能不是很清楚。输出（使用 T 方法）在第一行输出 { { { { { { { { { （每列一个），u u u u u u u u u u (Unicode) on next and then ' ' ' ' ' ' ' ' ' 下一个。
数据是否保密？