英文:
Pandas transform to datetime
问题
我已经收到一些以奇怪的格式给出的日期，我无法将它们解读为Pandas日期时间格式。
一个示例是：736698.0，它应该是'2017-01-04T00:00:00.000000000'。因此，原始格式似乎是自1/1/1BC以来的天数（作为零年的第一天，但没有这样的年份，所以是年份-1）。
我已经尝试使用pandas.to_datetime(736698.0, unit='D', origin=datetime.datetime(1/1/0))和其他组合，但我没有得到任何结果。
英文:
I've been given some dates in an strange format and I don't manage to read them as datetime pandas format.
An example would be: 736698.0, which should be '2017-01-04T00:00:00.000000000'. So, it seems the original format is days since 1/1/1BC (as first day of year zero, but there's no such year, so it's year -1).
I have tried to use pandas.to_datetime(736698.0, unit='D', origin=datetime.datetime(1/1/0)) and other combinations, but I'm not getting anything.
答案1
得分: 1
你可以使用以下方法将给定的浮点数转换为日期时间。由于无法设置year=0，我们必须从最终结果中减去它：
df['Date'] = df['Date_num'].apply(lambda x: datetime.datetime(1, 1, 1) + datetime.timedelta(x) - datetime.timedelta(366))
输出
Date_num Date
736698.0 2017-01-04
英文:
You can convert the given floats into a datetime with the following. As we cannot set a year=0, we must subtract that from the final result:
df['Date'] = df['Date_num'].apply(lambda x: datetime.datetime(1,1,1) + datetime.timedelta(x) - datetime.timedelta(366))
# Output
# Date_num Date
# 736698.0 2017-01-04
答案2
得分: 1
以下是您要翻译的内容：
如果您想要使用矢量方法并且期望最终日期是最近的，您可以减去一个有效的纪元：
# 使用有效的起始日期
epoch = pd.to_datetime('1970-1-1')
# 定义此起始日期相对于您的参考日期的值
epoch_from_ref = 719527
# 执行转换
df['date'] = pd.to_datetime(df['col'].sub(epoch_from_ref),
unit='D', origin=epoch)
输出：
col date
0 736698 2017-01-05
使用的输入：
df = pd.DataFrame({'col': [736698]})
英文:
If you want a vectorial method and expect the final dates to be recent, you can subtract a valid epoch:
# use a valid origin
epoch = pd.to_datetime('1970-1-1')
# define the value of this origin relative to your reference
epoch_from_ref = 719527
# perform the conversion
df['date'] = pd.to_datetime(df['col'].sub(epoch_from_ref),
unit='D', origin=epoch)
Output:
col date
0 736698 2017-01-05
Used input:
df = pd.DataFrame({'col': [736698]})
答案3
得分: 1
使用 numpy 可以支持‘BC’年份的操作：
import numpy as np
>>> np.datetime64('000') + np.timedelta64(736698, 'D')
numpy.datetime64('2017-01-04')
示例：
import pandas as pd
df = pd.DataFrame({'Date': [736698.0]})
df['Date2'] = pd.to_timedelta(df['Date'], unit='D').to_numpy() + np.datetime64('000')
输出：
>>> df
Date Date2
0 736698.0 2017-01-04
英文:
Use numpy to do that as it supports 'BC' years:
import numpy as np
>>> np.datetime64('000') + np.timedelta64(736698, 'D')
numpy.datetime64('2017-01-04')
Example:
import pandas as pd
df = pd.DataFrame({'Date': [736698.0]})
df['Date2'] = pd.to_timedelta(df['Date'], unit='D').to_numpy() + np.datetime64('000')
Output:
>>> df
Date Date2
0 736698.0 2017-01-04
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论