Timestamp subtraction of time arrays with different timezones

Question

I have the following code from somebody else that has a similar problem, but the solution proposed does not work on my DataFrame. The code subtracts a Pandas DataFrame index from a given date:

my_date = pd.datetime.today()
MyDF['day_differential'] = (MyDF.index - my_date).days

Which is generating the following error in my DataFrame:

TypeError: Timestamp subtraction must have the same timezones or no timezones

How do I found out tz for both dates? How do I make them the same so I can calculate the number of days between them?

Answer 1

Here is an answer using J.F. Sebastian's comment thanks really to him, because your index has timezone information then the operations must also be timezone aware, in your case the timezone is utc so you need to generate a utc timestamp to perform the subtraction:

In [11]:

import pandas as pd
import numpy as np
import datetime as dt
my_date = pd.datetime.today()
MyDF = pd.DataFrame({'a':np.random.randn(5)})
MyDF.index = pd.date_range('1/1/2011', periods=5, freq='H', tz='utc')
MyDF['day_differential'] = MyDF.index.tz_convert(None) - dt.datetime.utcnow()
MyDF
Out[11]:
                                  a            day_differential
2011-01-01 00:00:00+00:00  1.399602 -1493 days +13:04:06.875715
2011-01-01 01:00:00+00:00 -1.962517 -1493 days +14:04:06.875715
2011-01-01 02:00:00+00:00 -1.574531 -1493 days +15:04:06.875715
2011-01-01 03:00:00+00:00 -0.224702 -1493 days +16:04:06.875715
2011-01-01 04:00:00+00:00 -0.800772 -1493 days +17:04:06.875715

You can find out if your index is timezone aware by ouputting the index:

In [12]:

MyDF.index
Out[12]:
<class 'pandas.tseries.index.DatetimeIndex'>
[2011-01-01 00:00:00+00:00, ..., 2011-01-01 04:00:00+00:00]
Length: 5, Freq: H, Timezone: UTC

compare with a non timezone aware index:

In [14]:

MyDF.index
Out[14]:
<class 'pandas.tseries.index.DatetimeIndex'>
[2011-01-01 00:00:00, ..., 2011-01-01 04:00:00]
Length: 5, Freq: H, Timezone: None