Greetings, I would like to subset observations in a time series using xts, after converting from ts to xts. X=ts(1:100, frequency=12, start=c(1976)) X2=as.xts(X) X2["1984"] The output: Feb 1984 98 Mar 1984 99 Apr 1984 100 What happened to January? The index is always one month off, with X2["1976-01"] giving me Feb 1976. Should I set the time using something else than ts? I know there is the window function, but the xts format is convenient for my needs. Thank you. -- Philippe Baril Lecavalier
Time zones strike again! In short, your problem is that your machine is probably not in UTC for its timezone (and it probably shouldn't be) so 12:00am Jan1 in London is still Dec31 for you. To see this directly, look at as.POSIXct(time(X2)) -- you'll see that item 97 which "should be" Jan 1 is actually late in the day on Dec 31. Compare this to time(X2) and you'll get a nasty surprise. The right course of action is a little tricky: where is your data coming from? Is there a time zone associated with it and, if so, is it the same one your machine has? You can set the TZ of a xts object, but the kludgy (read: easier) way to do it might be to just add the offset to the time object: something like time(X2) <- time(X2) + 6*60*60. If you are doing monthly frequency stuff, this hopefully won't hurt anything but I make no promises. But if you can say more about the data source and your intended use, I'd be happy to give more particular thoughts. Hope this helps, Michael On Wed, Feb 15, 2012 at 10:25 PM, P.B. Lecavalier <p.barill at gmail.com> wrote:> Greetings, > > I would like to subset observations in a time series using xts, after > converting from ts to xts. > > X=ts(1:100, frequency=12, start=c(1976)) > X2=as.xts(X) > X2["1984"] > > The output: > > Feb 1984 ? 98 > Mar 1984 ? 99 > Apr 1984 ?100 > > What happened to January? The index is always one month off, with > X2["1976-01"] giving me Feb 1976. Should I set the time using something else > than ts? > > I know there is the window function, but the xts format is convenient for my > needs. > > Thank you. > > -- > Philippe Baril Lecavalier > > ______________________________________________ > 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.
On Wed, Feb 15, 2012 at 10:25 PM, P.B. Lecavalier <p.barill at gmail.com> wrote:> Greetings, > > I would like to subset observations in a time series using xts, after > converting from ts to xts. > > X=ts(1:100, frequency=12, start=c(1976)) > X2=as.xts(X) > X2["1984"] > > The output: > > Feb 1984 ? 98 > Mar 1984 ? 99 > Apr 1984 ?100 > > What happened to January? The index is always one month off, with > X2["1976-01"] giving me Feb 1976. Should I set the time using something else > than ts? > > I know there is the window function, but the xts format is convenient for my > needs.Any of these work with your sample data: X2[ trunc(time(X2)) == 1984 ] X2[ format(time(X2), "%Y") == "1984" ] window(X2, start = 1984) window(X, start = 1984) Sys.setenv(TZ = "GMT") X2["1984"] -- Statistics & Software Consulting GKX Group, GKX Associates Inc. tel: 1-877-GKX-GROUP email: ggrothendieck at gmail.com
Thank you for your replies. That can be quite tricky indeed.
The data consists of monthly unemployment rates, which I get as a text
file along with the information: monthly, period of beginning and end.
So it does not come with some timestamp as I specify the time inside R.
This tweak time(X2) <- time(X2) + 6*60*60 does not work. However this
solves the problem:
time(X2)=as.POSIXct(time(X2))+6*60*60
But then each element in my series now show up with time too instead of
only year/month alone.
However Sys.setenv(TZ = "GMT") works perfectly for this purpose.
Thank you for your insights!
--
Philippe Baril Lecavalier
On 02/15/2012 11:30 PM, Gabor Grothendieck wrote:> On Wed, Feb 15, 2012 at 10:25 PM, P.B. Lecavalier<p.barill at
gmail.com> wrote:
>> Greetings,
>>
>> I would like to subset observations in a time series using xts, after
>> converting from ts to xts.
>>
>> X=ts(1:100, frequency=12, start=c(1976))
>> X2=as.xts(X)
>> X2["1984"]
>>
>> The output:
>>
>> Feb 1984 98
>> Mar 1984 99
>> Apr 1984 100
>>
>> What happened to January? The index is always one month off, with
>> X2["1976-01"] giving me Feb 1976. Should I set the time using
something else
>> than ts?
>>
>> I know there is the window function, but the xts format is convenient
for my
>> needs.
> Any of these work with your sample data:
>
> X2[ trunc(time(X2)) == 1984 ]
>
> X2[ format(time(X2), "%Y") == "1984" ]
>
> window(X2, start = 1984)
>
> window(X, start = 1984)
>
> Sys.setenv(TZ = "GMT")
> X2["1984"]
>
>