[R] DateTime Math in R - POSIXct

Tue May 31 15:15:15 CEST 2011

Many thank, Jeff  – 

a)  I am indeed on Windoze, and found the time zone directory under: C:\Program Files\R\R-2.13.0\share\zoneinfo
b)  I applied Sys.setenv(TZ="America/Denver") to my real program, and the numerical and POSIXct output representations of my date/times now work, thanks! I've avoided converting back to numeric from POSIXct in my real program.  I left my calc'd variables in numeric mode so as to avoid POSIXct surprises.
c)  Oddly, however, my dummy/demo code below still shows some screwy behavior, even with Sys.setenv(TZ="America/Denver") applied.  I'm guessing that the apparent reversion to GMT occurs in my statement " dateN <- as.numeric(dateP)" below is merely the internal representation to which you refer.   I don't understand why POSIXct slaps another hour on dateN on conversion to dateP, nor why R generates the "MDT" timezone value (not listed and not valid tz designation).

Thanks again for your help, and grateful for any further hints.  I'd really like to be not confused by this approach to datetime arithmetic in R.

- Galen

> Sys.setenv(TZ="America/Denver")
> 
> tstamp <- "2011-05-22 11:45:00"
> mode(tstamp)
[1] "character"
> 
> dateP <- as.POSIXct(tstamp, origin="1970-01-01")
> mode(dateP)
[1] "numeric"
> dateP
[1] "2011-05-22 11:45:00 MDT"
> # Denver is currently on MDT, and note that R generates the "MDT" designation,
> # even though it is not listed in C:\Program Files\R\R-2.13.0\share\zoneinfo
> 
> dateN <- as.numeric(dateP)
> dateN
[1] 1306086300
> # 1306086300  returns "5/22/2011 17:45:00" when converted manually
> 
> dateP2 <- as.POSIXct(dateN, origin="1970-01-01")
> dateP2
[1] "2011-05-22 18:45:00 MDT"
> # Where does the extra hour come from?

From: Jeff Newmiller [mailto:jdnewmil at dcn.davis.ca.us] 
Sent: Tuesday, May 31, 2011 00:43
To: galen.a.moore at gmail.com; 'David Winsemius'
Cc: r-help at r-project.org
Subject: Re: [R] DateTime Math in R - POSIXct

MDT is not a valid timezone specification... it is falling back on your default timezone in your "working" case.

a) learn what valid timezone specifications are on your operating system. I suspect you are using Windows (hint: we should not have to guess), so look for the timezone directory in your R installation directory. I would guess that you want "America/Denver". Note that MST and MST7MDT are also possibilities , but they are generally deprecated as being too imprecise in the global context (who is to say "MDT" doesn't mean "Mongolian Daylight Time").

b) I strongly recommend using the "default" timezone or overriding it with Sys.setenv(TZ= your timezone spec) rather than using the tz arguments. The frustration factor is much lower.

c) if you properly import all of your time values into POSIXct, their internal representations will always be referenced to UTC (GMT). It is only on input and output (such as printing to the console) or conversion to POSIXlt that the timezone is considered.
---------------------------------------------------------------------------
Jeff Newmiller The ..... ..... Go Live...
DCN:<jdnewmil at dcn.davis.ca.us> Basics: ##.#. ##.#. Live Go...
Live: OO#.. Dead: OO#.. Playing
Research Engineer (Solar/Batteries O.O#. #.O#. with
/Software/Embedded Controllers) .OO#. .OO#. rocks...1k
--------------------------------------------------------------------------- 
Sent from my phone. Please excuse my brevity.
Galen Moore <galen.a.moore at gmail.com> wrote:
Thank you, David.

I am not subtracting the seconds to convert POSIXct's GMT to MDT, and don't
understand why I should need to.

Any hints, however, as to why  
dateP <- as.POSIXct(tstamp, origin="1970-01-01", tzone="MDT")  
returns a date in the correct tz in my first instance below, yet 

dateP2 <- as.POSIXct(dateN, origin="1970-01-01", tzone="MDT")  
decides to revert to GMT in my second instance below?

Thanks,

Galen

-----Original Message-----
From: David Winsemius [mailto:dwinsemius at comcast.net] 
Sent: Monday, May 30, 2011 20:31
To: galen.a.moore at gmail.com
Cc: r-help at r-project.org
Subject: Re: [R] DateTime Math in R - POSIXct

On May 30, 2011, at 10:20 PM, Galen Moore wrote:

> Greetings -
>
>
>
> I'm battling
 g
POSIXct, as per the code below.  My input is actually an 
> XL file, but the weird results below correctly model what I am seeing 
> in my program.
>
>
>
> Before I punt and use lubridate or timeDate, could anyone please help 
> me understand why POSIXct forces my variable back to GMT?
>
>
>
> I suspect that I'm not properly coding the tzone value, but it does 
> not throw an error as-is.
>
>
>
>
>
>> tstamp <- "2011-05-22 11:45:00 MDT"
>
>> mode(tstamp)
>
> [1] "character"
>
>>
>
>> dateP <- as.POSIXct(tstamp, origin="1970-01-01", tzone="MDT")
>
>> mode(dateP)
>
> [1] "numeric"
>
>> dateP
>
> [1] "2011-05-22 11:45:00 MDT"
>
>>
>
>> dateN <-
as.numeric(dateP)
>
>> dateN
>
> [1] 1306086300
>

So now the internal representation is referenced to GMT

>>
>
>> dateP2 <- as.POSIXct(dateN, origin="1970-01-01", tzone="MDT")
>
>> dateP2

And you are telling the function to add as many hours offset as needed for
the difference between GMT and MD T.....

--
David.
>
> [1] "2011-05-22 18:45:00 MDT"
>

David Winsemius, MD
West Hartford, CT
________________________________________

R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.