[R] ts instead of xts object

[Joshua Ulrich]

On Sun, Mar 16, 2014 at 6:27 AM, Bill <william108 at gmail.com> wrote:
> Hello Pascal,
> Yes that is what I was worried about. The date-stamps are there and I would
> like to use that information but I think using as.ts will not do this.
> Does anyone know how this is done?

It cannot be done. The ts class is used to "represent data which has
been sampled at equispaced points in time."  You need to look at the
changepoint package code (or ask the author(s)) to determine whether
or not the frequency attribute is used.

> Thank you.
>
>
> On Wed, Mar 12, 2014 at 1:02 AM, Pascal Oettli <kridox at ymail.com> wrote:
>>
>> Hello,
>>
>> On Tue, Mar 11, 2014 at 8:45 PM, Joshua Ulrich <josh.m.ulrich at gmail.com>
>> wrote:
>> > On Tue, Mar 11, 2014 at 12:14 AM, Bill <william108 at gmail.com> wrote:
>> >>
>> >> Hello. I have a dataframe that has a date column. The intervals between
>> >> dates vary. I want to convert this to a ts object. I was able to
>> >> convert it
>> >> to an xts object but the package I want to analyse this data with
>> >> (called
>> >> 'changepoint') does not seem to want to deal with xts. In the example
>> >> they
>> >> give they use the following:
>> >>
>> >> data(discoveries)
>> >> dis.pelt=cpt.meanvar(discoveries,test.stat='Poisson',method='PELT')
>> >> plot(dis.pelt,cpt.width=3)
>> >> cpts.ts(dis.pelt)
>> >>
>> >> and if I check:
>> >> str(discoveries)
>> >>  Time-Series [1:100] from 1860 to 1959: 5 3 0 2 0 3 2 3 6 1 ...
>> >>
>> >> If I try with my data
>> >> str(testTSRad)
>> >> An 'xts' object on 2011-07-16 07:08:02/2013-09-20 01:25:48 containing:
>> >>   Data: num [1:501, 1] 76 77 79 86 79 79 85 86 89 88 ...
>> >>   Indexed by objects of class: [POSIXct,POSIXt] TZ:
>> >>   xts Attributes:
>> >>  NULL
>> >>
>> >> where I used this:
>> >>
>> >> testTSRad=xts(radSampPerRegion[[2]][
>> >> ,2],order.by=as.POSIXct(radSampPerRegion[[2]][
>> >> ,1]))
>> >>
>> >> I get this:
>> >>
>> >> testt=cpt.mean(testTSRad)
>> >> Error in single.mean.norm(data, penalty, pen.value, class,
>> >> param.estimates)
>> >> :
>> >>   Data must have atleast 2 observations to fit a changepoint model.
>> >>
>> > This is because of what ?cpt.mean says about the "data" argument:
>> > data: A vector, ts object or matrix containing the data within
>> >       which you wish to find a changepoint.  If data is a matrix,
>> >       each row is considered a separate dataset.
>> >
>> > An xts object is a matrix (with an index attribute), so each row is
>> > considered a separate data set.  Your object only has one column,
>> > hence only one observation per data set.  Things will work if you drop
>> > the dimensions of your single-column xts object:
>> > testt <- cpt.mean(drop(testTSRad))
>> >
>> >> My data is below. Is there a way to convert it to ts?
>> >>
>> > Yes, as is generally the case, use the "as" method:
>> > as.ts(testTSRad)
>> >
>>
>> But in this case, the time serie will have a frequency of 1, which is
>> inconsistent with irregular sampling. This probably will lead to
>> inaccurate results
>>
>> > Best,
>> > --
>> > Joshua Ulrich  |  about.me/joshuaulrich
>> > FOSS Trading  |  www.fosstrading.com
>> >
>> > ______________________________________________
>> > 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.
>>
>> Regards,
>> Pascal
>>
>> --
>> Pascal Oettli
>> Project Scientist
>> JAMSTEC
>> Yokohama, Japan
>
>