熊猫从UTC到当地时间

Question

我正在尝试将UTC时间转换为当地时间。这是我以前的经历

df_combined_features['timestamp'][1:10]
2013-01-24   2013-01-24 11:00:00
2013-04-25   2013-04-25 10:00:00
2013-07-25   2013-07-25 10:00:00
2013-10-24   2013-10-24 10:00:00
2014-01-30   2014-01-30 11:00:00
2014-04-24   2014-04-24 10:00:00
2014-07-24   2014-07-24 10:00:00
2014-10-23   2014-10-23 10:00:00
2015-01-27   2015-01-27 11:00:00

这就是我所做的

df_combined_features['time_stamp'].tz_localize('US/Central')[1:10]
2013-01-24 00:00:00-06:00   2013-01-24 11:00:00
2013-04-25 00:00:00-05:00   2013-04-25 10:00:00
2013-07-25 00:00:00-05:00   2013-07-25 10:00:00
2013-10-24 00:00:00-05:00   2013-10-24 10:00:00
2014-01-30 00:00:00-06:00   2014-01-30 11:00:00
2014-04-24 00:00:00-05:00   2014-04-24 10:00:00
2014-07-24 00:00:00-05:00   2014-07-24 10:00:00
2014-10-23 00:00:00-05:00   2014-10-23 10:00:00
2015-01-27 00:00:00-06:00   2015-01-27 11:00:00

我认为它做对了，但我不理解输出格式。特别是

1）为什么转换后的列显示为新索引？

2）我知道-06：00（在最后一行）是一个小时班，所以时间是6:00，我该如何检索该信息以及确切的当地时间？

所需的输出，我希望发布确切的时间，包括与utc的偏移量。世界标准时间

    2013-01-24 05:00:00   2013-01-24 11:00:00
    2013-04-25 05:00:00   2013-04-25 10:00:00
    2013-07-25 05:00:00   2013-07-25 10:00:00
    2013-10-24 05:00:00   2013-10-24 10:00:00
    2014-01-30 05:00:00   2014-01-30 11:00:00
    2014-04-24 05:00:00   2014-04-24 10:00:00
    2014-07-24 05:00:00   2014-07-24 10:00:00
    2014-10-23 05:00:00   2014-10-23 10:00:00
    2015-01-27 05:00:00   2015-01-27 11:00:00

Answer 1

调用时tz.localize，对索引进行本地化，如果要修改列，则还需要调用dt.localize以添加时区偏移量调用dt.tz_convert('UTC')：

In [125]:
df['timestamp'].dt.tz_localize('utc').dt.tz_convert('US/Central')

Out[125]:
index
2013-01-24   2013-01-24 05:00:00-06:00
2013-04-25   2013-04-25 05:00:00-05:00
2013-07-25   2013-07-25 05:00:00-05:00
2013-10-24   2013-10-24 05:00:00-05:00
2014-01-30   2014-01-30 05:00:00-06:00
2014-04-24   2014-04-24 05:00:00-05:00
2014-07-24   2014-07-24 05:00:00-05:00
2014-10-23   2014-10-23 05:00:00-05:00
2015-01-27   2015-01-27 05:00:00-06:00
Name: timestamp, dtype: datetime64[ns, US/Central]

不比较.dt：

In [126]:    
df['timestamp'].tz_localize('utc').tz_convert('US/Central')
Out[126]:
index
2013-01-23 18:00:00-06:00   2013-01-24 11:00:00
2013-04-24 19:00:00-05:00   2013-04-25 10:00:00
2013-07-24 19:00:00-05:00   2013-07-25 10:00:00
2013-10-23 19:00:00-05:00   2013-10-24 10:00:00
2014-01-29 18:00:00-06:00   2014-01-30 11:00:00
2014-04-23 19:00:00-05:00   2014-04-24 10:00:00
2014-07-23 19:00:00-05:00   2014-07-24 10:00:00
2014-10-22 19:00:00-05:00   2014-10-23 10:00:00
2015-01-26 18:00:00-06:00   2015-01-27 11:00:00
Name: timestamp, dtype: datetime64[ns]