Datetime module and Pandas to_datetime give different results

问题:

I have a string containing a UTC datetime

utc_str = '2017-11-21T23:00+0100' 

which in my local time (Europe/Berlin) is:

local_time = '2017-11-22 00:00' 

And is the desired value I would like to obtain from utc_string.

I can convert utc_string to local_time just fine using:

import datetime as dt utc_time = dt.datetime.strptime(date_str, '%Y-%m-%dT%H:%M%z') local_time = utc_time.replace(tzinfo=pytz.utc).astimezone(pytz.timezone('Europe/Berlin'))  print(local_time.strftime('%Y-%m-%d %H:%M')) >>> 2017-11-22 00:00 

However, when I use Pandas, I get a different result. It doesn't seem to apply the UTC offset:

import pandas as pd pd_date = pd.to_datetime(date_str, utc=True)  print(pd_date.strftime('%Y-%m-%d %H:%M')) >>> '2017-11-21 22:00' 

And naively if I try to do the same process as with the datetime module, the results are still off:

pd_date = pd.to_datetime(date_str, utc=True) pd_date = pd_date.replace(tzinfo=pytz.utc).astimezone(pytz.timezone('Europe/Berlin'))  print(pd_date.strftime('%Y-%m-%d %H:%M')) >>> '2017-11-21 23:00' 

Is there something I am not understanding? Am I using pd.to_datetime or something else wrong? On Python 3.6, Windows 7.

回答1:

As stated in the comment, I think your code for local_time is wrong

utc_time 

datetime.datetime(2017, 11, 21, 23, 0, tzinfo=datetime.timezone(datetime.timedelta(0, 3600)) 

utc_time.replace(tzinfo=pytz.utc) 

'datetime.datetime(2017, 11, 21, 23, 0, tzinfo=<UTC>)' 

so this replace removes the '+0100 from the datetime, but keeps the rest the same

utc_time.replace(tzinfo=pytz.utc).astimezone(pytz.timezone('Europe/Berlin')) 

"datetime.datetime(2017, 11, 22, 0, 0, tzinfo=<DstTzInfo 'Europe/Berlin' CET+1:00:00 STD>)" 

This then adds 1 hour to 23:00UTC, so become the next day midnight in Berlin as expected

pd.to_datetime(utc_str, utc=True) 

Timestamp('2017-11-21 22:00:00+0000', tz='UTC') 

The difference in behaviour is due to the constructor. pd.to_datetime calculates the time and timezone back to 22:00UTC instead of 23:00+0100, so if there you replace the timezone info with UTC, it changes nothing

Local time

Your utc_time object is in the correct timezone, so if you want the local time you can just do utc_time.strftime('%Y-%m-%d %H:%M') in pandas you'll have to do pd.to_datetime(utc_str, utc=True).astimezone(pytz.timezone('Europe/Berlin')).strftime('%Y-%m-%d %H:%M')

文章来源: Datetime module and Pandas to_datetime give different results