Pandas date_range 非预期行为

huangapple go评论70阅读模式
英文:

Pandas date_range unexpected behaviour

问题

如果我运行:

pd.date_range('2022-01-01 15:00', '2023-02-02 00:00', freq='YS')

我会收到:

DatetimeIndex(['2022-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

我期望在返回中也会得到'2023-01-01 15:00:00',即:

DatetimeIndex(['2022-01-01 15:00:00', '2023-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

我觉得这很奇怪。我是否遗漏了什么,还是这是一个错误?如何实现预期的行为?

谢谢

编辑:同意s_pike的观点,这种行为看起来相当反直觉,我认为这是一个错误。

英文:

If I run:

pd.date_range('2022-01-01 15:00', '2023-02-02 00:00', freq='YS')

I receive:

DatetimeIndex(['2022-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

I would expect to get 2023-01-01 15:00:00 in the return as well, i.e.

DatetimeIndex(['2022-01-01 15:00:00', '2023-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

Feel this is really strange. Am I missing anything or is it a bug? How may I achieve the expected behaviour?

Thanks

Edit: agree with s_pike, this behaviour looks quite counter-intuitive and I would say it's a bug.

答案1

得分: 0

你可以尝试这样做,这可能会提供您期望的结果。

date_range = pd.date_range('2022-01-01 15:00', '2023-02-02 00:00', freq='AS')
date_range = date_range.union(pd.DatetimeIndex(['2023-01-01 15:00']))

这个pd.DatetimeIndex()构造函数将创建一个新的"datetimeindex"。

英文:

You can try like this, this might provide your expected result.

date_range = pd.date_range('2022-01-01 15:00', '2023-02-02 00:00', freq='AS')
date_range = date_range.union(pd.DatetimeIndex(['2023-01-01 15:00']))

This pd.DatetimeIndex() constructor will create a new "datetimeindex".

答案2

得分: 0

以下是翻译好的内容:

你可以通过以下方式获取你想要的输出:

pd.date_range('2022-01-01 15:00', '2023-01-01 15:00', freq='YS')
    
# DatetimeIndex(['2022-01-01 15:00:00', '2023-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

如果你将日期更改为二月,你必须将时间更改为大于或等于开始日期中给定的时间。

pd.date_range('2022-01-01 15:00', '2023-02-02 15:00', freq='YS')
    
# DatetimeIndex(['2022-01-01 15:00:00', '2023-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

如果你的时间少于开始日期中指定的时间,那么你将无法获得结果:

pd.date_range('2022-01-01 15:00', '2023-02-02 14:59', freq='YS')
    
# DatetimeIndex(['2022-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

以下是相关文档链接:

pandas.date_range文档

pandas时间序列指标别名文档

英文:

You can get your desired output by:

pd.date_range('2022-01-01 15:00', '2023-01-01 15:00', freq='YS')

#DatetimeIndex(['2022-01-01 15:00:00', '2023-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

If you change your date to February you must change time to greater than or equal time given in start date.

pd.date_range('2022-01-01 15:00', '2023-02-02 15:00', freq='YS')

#DatetimeIndex(['2022-01-01 15:00:00', '2023-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

If you less time than the time specified in the start date then you will not get the result:

pd.date_range('2022-01-01 15:00', '2023-02-02 14:59', freq='YS')

#DatetimeIndex(['2022-01-01 15:00:00'], dtype='datetime64[ns]', freq='AS-JAN')

here are the links to documents:

https://pandas.pydata.org/docs/reference/api/pandas.date_range.html

https://pandas.pydata.org/docs/user_guide/timeseries.html#timeseries-offset-aliases

huangapple
  • 本文由 发表于 2023年3月8日 19:14:46
  • 转载请务必保留本文链接:https://go.coder-hub.com/75672291.html
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定