在熊猫中将索引转换为日期时间答案

【问题标题】：Convert index to datetime in pandas在熊猫中将索引转换为日期时间
【发布时间】：2019-01-15 17:24:58
【问题描述】：

我正在尝试使用 to_datetime() 方法将 Pandas 数据帧的索引转换为日期时间，但出现异常。

对于一个玩具可重现的例子：

我的数据：

json格式：

'{"0":{"Id":1,"Name":"Lewis Alexander","Organization":"Nomura Securities International","Dec 2018":3.25,"June 2019":3.25,"Dec 2019":3.0,"June 2020":3.0,"Dec 2020":2.88},"1":{"Id":2,"Name":"Scott Anderson","Organization":"Bank of the West","Dec 2018":3.19,"June 2019":3.5,"Dec 2019":3.47,"June 2020":3.1,"Dec 2020":2.6},"2":{"Id":3,"Name":"Paul Ashworth","Organization":"Capital Economics","Dec 2018":3.25,"June 2019":3.0,"Dec 2019":2.5,"June 2020":2.25,"Dec 2020":2.25},"3":{"Id":4,"Name":"Daniel Bachman","Organization":"Deloitte LP","Dec 2018":3.2,"June 2019":3.4,"Dec 2019":3.5,"June 2020":3.6,"Dec 2020":3.7},"4":{"Id":5,"Name":"Bernard Baumohl","Organization":"Economic Outlook Group","Dec 2018":3.1,"June 2019":3.35,"Dec 2019":3.6,"June 2020":3.9,"Dec 2020":4.2}}'

当我尝试时：

df3.index.to_datetime()

我收到以下错误消息：

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-53-5ff789e24651> in <module>()
----> 1 df3.index.to_datetime()

AttributeError: 'Index' object has no attribute 'to_datetime'

【问题讨论】：

to_datetime 是一个顶级函数，它不是 Index 对象的方法，所以它应该是 df3.index = pd.to_datetime(df3.index)
OP 有两个故障点。 @EdChum 确定了一个。另一个是当前索引不适合成为DatetimeIndex，这也需要解决。

标签： python pandas datetime indexing

【解决方案1】：

您不能拥有包含非日期时间内容的 DatetimeIndex。我的建议是将标记为['Id', 'Name', 'Organization'] 的行放入MultiIndex 列对象中。

df = df.T.set_index(['Id', 'Name', 'Organization']).T
df = df.set_index(pd.to_datetime(df.index))
df

Id                                         1                2                 3              4                      5
Name                         Lewis Alexander   Scott Anderson     Paul Ashworth Daniel Bachman        Bernard Baumohl
Organization Nomura Securities International Bank of the West Capital Economics    Deloitte LP Economic Outlook Group
2018-12-01                              3.25             3.19              3.25            3.2                    3.1
2019-12-01                                 3             3.47               2.5            3.5                    3.6
2020-12-01                              2.88              2.6              2.25            3.7                    4.2
2019-06-01                              3.25              3.5                 3            3.4                   3.35
2020-06-01                                 3              3.1              2.25            3.6                    3.9

您可以创建一个混合类型的索引的可憎。如果不是很明显，我真的建议不要这样做。如果您足够关心索引的类型并尝试将其转换为Datetime，那么您不应该这样做。

df.set_index(df.index.map(lambda x: pd.to_datetime(x, errors='ignore')))

                                                   0                 1                  2               3                       4
2018-12-01 00:00:00                             3.25              3.19               3.25             3.2                     3.1
2019-12-01 00:00:00                                3              3.47                2.5             3.5                     3.6
2020-12-01 00:00:00                             2.88               2.6               2.25             3.7                     4.2
Id                                                 1                 2                  3               4                       5
2019-06-01 00:00:00                             3.25               3.5                  3             3.4                    3.35
2020-06-01 00:00:00                                3               3.1               2.25             3.6                     3.9
Name                                 Lewis Alexander    Scott Anderson      Paul Ashworth  Daniel Bachman         Bernard Baumohl
Organization         Nomura Securities International  Bank of the West  Capital Economics     Deloitte LP  Economic Outlook Group

【讨论】：