问题：
我正在尝试使用日期，时间和时区列创建DateTime列。时区是异构的。随后将新创建的DateTime列中的时间转换为本地TimeZone。下面我用示例代码解释了我的用例。
用例和示例代码：
我有两个pandas DataFrame，即df1和df2。
df1在本地TimeZone中为不同国家/地区的学校提供开放/关闭时间，并提供各自的时区。< / p>
>>> df1 = pd.DataFrame({'School': {0: 'ABC', 2: 'GHI', 3: 'JKL', 4: 'MNO'}, 'OpenTime': {0: '08:00:00.000', 2: '10:00:23.563', 3: '09:30.05.908', 4: '07:15:50.100'}, 'CloseTime': {0: '13:00:00.000', 2: '13:30:00.100', 3: '15:00.00.768', 4: '13:00:00.500'}, 'TimeZone':{0:'Europe/Vienna', 2:'Europe/London', 3:'Pacific/Auckland', 4:'Asia/Seoul'}})
>>> df1
CloseTime OpenTime School TimeZone
0 13:00:00.000 08:00:00.000 ABC Europe/Vienna
2 13:30:00.100 10:00:23.563 GHI Europe/London
3 15:00.00.768 09:30.05.908 JKL Pacific/Auckland
4 13:00:00.500 07:15:50.100 MNO Asia/Seoul
df2只是一个带有大量日期的中间数据框。
>>> df2 = pd.DataFrame({'Dates': {0: '2016-11-02', 1: '2015-03-31', 2: '2015-10-30', 3: '2001-09-01'}})
>>> df2
Dates
0 2016-11-02
1 2015-03-31
2 2015-10-30
3 2001-09-01
df3只是df1和df2的笛卡尔积，如下所示：
>>> df1['key'], df2['key'] =0,0
>>> df3 = df1.merge(df2, how='left', on='key')
>>> df3
CloseTime OpenTime School TimeZone key Dates
0 13:00:00.000 08:00:00.000 ABC Europe/Vienna 0 2016-11-02
1 13:00:00.000 08:00:00.000 ABC Europe/Vienna 0 2016-05-02
2 13:00:00.000 08:00:00.000 ABC Europe/Vienna 0 2015-03-31
...
现在我想找出在df3上执行以下两个步骤的最佳方法：1）使用{{1}将OpenTime和CloseTime列转换为DateTime }和Dates信息。
2）将TimeZone和OpenTime转换为'Europe / London'TimeZone。
答案 0 :(得分：0)
我认为您首先需要to_datetime然后apply才能转换为不同的时区。上次使用tz_localize首先重置为UTC，然后本地化为Europe/London：
df1 = pd.DataFrame({'School': {0: 'ABC', 2: 'GHI', 3: 'JKL', 4: 'MNO'}, 'OpenTime': {0: '08:00:00.000', 2: '10:00:23.563', 3: '09:30:05.908', 4: '07:15:50.100'}, 'CloseTime': {0: '13:00:00.000', 2: '13:30:00.100', 3: '15:00:00.768', 4: '13:00:00.500'}, 'TimeZone':{0:'Europe/Vienna', 2:'Europe/London', 3:'Pacific/Auckland', 4:'Asia/Seoul'}})
df2 = pd.DataFrame({'Dates': {0: '2016-11-02', 1: '2015-03-31', 2: '2015-10-30', 3: '2001-09-01'}})
df1['key'], df2['key'] = 0,0
df3 = df1.merge(df2, how='left', on='key')
print (df3)
df3['Close'] = pd.to_datetime(df3.Dates +' '+ df3.CloseTime)
df3['Open'] = pd.to_datetime(df3.Dates +' '+ df3.OpenTime)
df3['CloseFin'] = df3.apply(lambda x: x.Close.tz_localize(x.TimeZone), axis=1)
df3['OpenFin'] = df3.apply(lambda x: x.Open.tz_localize(x.TimeZone), axis=1)
df3['OpenEULondon'] = df3['Open'].dt.tz_localize('UTC').dt.tz_convert('Europe/London')
df3['CloseEULondon'] = df3['Close'].dt.tz_localize('UTC').dt.tz_convert('Europe/London')
print (df3)
...
...
CloseFin OpenFin \
0 2016-11-02 13:00:00+01:00 2016-11-02 08:00:00+01:00
1 2015-03-31 13:00:00+02:00 2015-03-31 08:00:00+02:00
2 2015-10-30 13:00:00+01:00 2015-10-30 08:00:00+01:00
3 2001-09-01 13:00:00+02:00 2001-09-01 08:00:00+02:00
4 2016-11-02 13:30:00.100000+00:00 2016-11-02 10:00:23.563000+00:00
5 2015-03-31 13:30:00.100000+01:00 2015-03-31 10:00:23.563000+01:00
6 2015-10-30 13:30:00.100000+00:00 2015-10-30 10:00:23.563000+00:00
7 2001-09-01 13:30:00.100000+01:00 2001-09-01 10:00:23.563000+01:00
8 2016-11-02 15:00:00.768000+13:00 2016-11-02 09:30:05.908000+13:00
9 2015-03-31 15:00:00.768000+13:00 2015-03-31 09:30:05.908000+13:00
10 2015-10-30 15:00:00.768000+13:00 2015-10-30 09:30:05.908000+13:00
11 2001-09-01 15:00:00.768000+12:00 2001-09-01 09:30:05.908000+12:00
12 2016-11-02 13:00:00.500000+09:00 2016-11-02 07:15:50.100000+09:00
13 2015-03-31 13:00:00.500000+09:00 2015-03-31 07:15:50.100000+09:00
14 2015-10-30 13:00:00.500000+09:00 2015-10-30 07:15:50.100000+09:00
15 2001-09-01 13:00:00.500000+09:00 2001-09-01 07:15:50.100000+09:00
OpenEULondon CloseEULondon
0 2016-11-02 08:00:00+00:00 2016-11-02 13:00:00+00:00
1 2015-03-31 09:00:00+01:00 2015-03-31 14:00:00+01:00
2 2015-10-30 08:00:00+00:00 2015-10-30 13:00:00+00:00
3 2001-09-01 09:00:00+01:00 2001-09-01 14:00:00+01:00
4 2016-11-02 10:00:23.563000+00:00 2016-11-02 13:30:00.100000+00:00
5 2015-03-31 11:00:23.563000+01:00 2015-03-31 14:30:00.100000+01:00
6 2015-10-30 10:00:23.563000+00:00 2015-10-30 13:30:00.100000+00:00
7 2001-09-01 11:00:23.563000+01:00 2001-09-01 14:30:00.100000+01:00
8 2016-11-02 09:30:05.908000+00:00 2016-11-02 15:00:00.768000+00:00
9 2015-03-31 10:30:05.908000+01:00 2015-03-31 16:00:00.768000+01:00
10 2015-10-30 09:30:05.908000+00:00 2015-10-30 15:00:00.768000+00:00
11 2001-09-01 10:30:05.908000+01:00 2001-09-01 16:00:00.768000+01:00
12 2016-11-02 07:15:50.100000+00:00 2016-11-02 13:00:00.500000+00:00
13 2015-03-31 08:15:50.100000+01:00 2015-03-31 14:00:00.500000+01:00
14 2015-10-30 07:15:50.100000+00:00 2015-10-30 13:00:00.500000+00:00
15 2001-09-01 08:15:50.100000+01:00 2001-09-01 14:00:00.500000+01:00
但dtypes列和CloseFin OpenFin为object，see：
print (df3.dtypes)
CloseTime object
OpenTime object
School object
TimeZone object
key int64
Dates object
Close datetime64[ns]
Open datetime64[ns]
CloseFin object
OpenFin object
OpenEULondon datetime64[ns, Europe/London]
CloseEULondon datetime64[ns, Europe/London]
dtype: object