CODEWITHSUNDEEP
pythonpandastime
user62198
user62198

Reputation: 1874

confusion with converting epoch time to local time in Python

I have epoch time which I need to convert to local time (US/Pacific). I get the following results using pandas to_datetime() function and datetime.datetime.fromtimestamp().

datetime.datetime.fromtimestamp(1554782410.0370002)
Out[704]: datetime.datetime(2019, 4, 8, 21, 0, 10, 37000)

pd.to_datetime(1554782410.0370002, unit='s')
Out[705]: Timestamp('2019-04-09 04:00:10.037000179')

Note that the time difference is 7:00 hrs between the two output. How do I reconcile the two?

Any help would be greatly appreciated.

Thanks.

Upvotes: 0

Views: 50

Answers (2)

anon01
anon01

Reputation: 11171

I prefer to define in UTC (tz aware) then convert:

pd.to_datetime(1554782410.0370002, unit='s', utc=True).tz_convert('US/Pacific')

The obvious thing you dont want to do manually subtract hours. If you happen to move, or your code runs on a server in another timezone, it will fail if you manually subtract off the hours. In short: use a package/api for timezone management if you want timezone-aware times.

Upvotes: 1

goalie1998
goalie1998

Reputation: 1432

Two options as I see it:

First is to manually subtract 7 hours using pd.Timedelta():

pd.to_datetime(1554782410.0370002, unit='s') - pd.Timedelta(hours=7)

Second would be to convert it to local time zone from UTC time using tz_localize and tz_convert. I think this method is slower, but does have the advantage of not worrying about calculating time difference manually or worrying about DST:

pd.to_datetime(1554782410.0370002, unit='s', utc=True).tz_convert('America/Los_Angeles')

Upvotes: 0

Related Questions