[R] Error message when adding drift for Arima model
Michael Howell
mchowe||2 @end|ng |rom gm@||@com
Mon May 13 20:35:46 CEST 2019
Yes that is a little off topic but I will look into it more. Thank you very
much for your help.
Michael
On Mon, May 13, 2019 at 11:33 AM Rui Barradas <ruipbarradas using sapo.pt> wrote:
> Hello,
>
> Sorry for the late reply.
> Inline.
>
> Às 17:54 de 10/05/19, Michael Howell escreveu:
> > Rui,
> > I'm still new to ARIMA forecasting but I examined the PACF and saw
> > significant correlation at lag 2.
>
> You saw a PACF with a significant correlation at lag 2 but not at lag 1.
> When this happens, it many times means that you shouldn't consider the
> lag 2. In fact, it might mean that the process is nonlinear.
> And the ACF shows an insignificant lag 1.
>
> Try
>
> ords <- list(c(1, 0, 0), c(2, 0, 0), c(0, 0, 1))
> fit_list <- lapply(ords, function(o)
> Arima(tsdata, order = o, include.drift = TRUE))
> sapply(fit_list, AIC)
> sapply(fit_list, BIC)
>
>
> Which gives the minimum AIC? And BIC?
> These are not perfect and automated model selection can have problems,
> but it's not unreasonable to compare them.
>
> I believe this is off-topic for R-Help, since it's a question about
> statistics and nonlinear time series is a really, really broad field to
> be discussed here. Try to find local help on this.
>
> Hope this helps,
>
> Rui Barradas
>
> The ACF showed a more gradual decline
> > which seemed to indicate it was Autoregressive. That should mean it's a
> > AR(2) process right?
> >
> > image.png
> > **//___^
> > Regards,
> > Michael Howell
> >
> >
> > On Fri, May 10, 2019 at 12:51 AM Rui Barradas <ruipbarradas using sapo.pt
> > <mailto:ruipbarradas using sapo.pt>> wrote:
> >
> > Why not
> >
> > Arima(tsdata, c(0, 0, 1), include.drift = TRUE)
> >
> > ?
> >
> > Why do you say it should be an AR(2) model?
> >
> > Hope this helps,
> >
> > Rui Barradas
> >
> > Às 06:43 de 10/05/19, Rui Barradas escreveu:
> > > Hello,
> > >
> > > This is just a typo, in R logical values ("true) are not character
> > > strings. You must pass FALSE (the default, can be omited) or TRUE.
> > >
> > > fitdata <- Arima(tsdata, c(2, 0, 0), include.drift = TRUE)
> > >
> > >
> > > From the help page ?logical
> > >
> > > Details
> > >
> > > TRUE and FALSE are reserved words denoting logical constants in
> > the R
> > > language, whereas T and F are global variables whose initial
> > values set
> > > to these. All four are logical(1) vectors.
> > >
> > > Hope this helps,
> > >
> > > Rui Barradas
> > >
> > > Às 00:26 de 10/05/19, Bert Gunter escreveu:
> > >> In future, always cc the list (unless it's personal,which this
> > isn't). I
> > >> have done so here. As I am largely ignorant on the subject
> > matter, others
> > >> will have to help, which is why you should cc the list.
> > >>
> > >> Cheers,
> > >> Bert Gunter
> > >>
> > >> "The trouble with having an open mind is that people keep coming
> > along
> > >> and
> > >> sticking things into it."
> > >> -- Opus (aka Berkeley Breathed in his "Bloom County" comic strip
> )
> > >>
> > >>
> > >> On Thu, May 9, 2019 at 3:49 PM Michael Howell
> > <mchowell2 using gmail.com <mailto:mchowell2 using gmail.com>>
> > >> wrote:
> > >>
> > >>> I apologize for that. The Arima() function that I'm trying to
> > use comes
> > >>> from the forecast package. I created a time series object using
> > the
> > >>> above
> > >>> 24 observations. The initial model I created doesn't seem to
> > perform so
> > >>> well so I thought a drift term might fit the data better. I
> > used the
> > >>> following code to create the time series object:
> > >>>
> > >>> tsdata<- ts(data, start = c(1,1), end = c(24,1), frequency = 1)
> > >>>
> > >>>
> > >>> Where* data* is the dataframe that contains the initial 24
> > observations.
> > >>> I then used the following code to try to create the model:
> > >>>
> > >>> fitdata <- Arima(tsdata,c(2,0,0),include.drift="true")
> > >>>>
> > >>>
> > >>> After doing this I obtained the following error message:
> > >>>
> > >>> Error in (order[2] + seasonal$order[2]) > 1 & include.drift:
> > operations
> > >>>> are possible only for numeric, logical or complex types
> > >>>> Traceback:
> > >>>>
> > >>>> 1. Arima(tsdata, c(2, 0, 0), include.drift = "true")
> > >>>
> > >>>
> > >>> I hope this is more clear.
> > >>>
> > >>> On Thu, May 9, 2019 at 4:39 PM Bert Gunter
> > <bgunter.4567 using gmail.com <mailto:bgunter.4567 using gmail.com>>
> > >>> wrote:
> > >>>
> > >>>> Please start by reading and following the posting guide linked
> > at the
> > >>>> bottom of this email. In particular:
> > >>>>
> > >>>> 1) Post in **plain text** on this plain text list so we don't
> > get the
> > >>>> mangled html of your post.
> > >>>>
> > >>>> 2) Tell us what package Arima() is in.
> > >>>>
> > >>>> Cheers,
> > >>>> Bert Gunter
> > >>>>
> > >>>>
> > >>>>
> > >>>>
> > >>>> On Thu, May 9, 2019 at 2:27 PM Michael Howell
> > <mchowell2 using gmail.com <mailto:mchowell2 using gmail.com>>
> > >>>> wrote:
> > >>>>
> > >>>>> Hello everyone,
> > >>>>> So this is my first post to this list, I'm trying to fit an
> > Arima
> > >>>>> (2,0,0)
> > >>>>> model and I think a drift term would help but I'm getting an
> > error
> > >>>>> term
> > >>>>> when I'm trying to include it. Here is my data:
> > >>>>>
> > >>>>> -6.732172338
> > >>>>> -2.868884273
> > >>>>> -5.371585089
> > >>>>> -6.512740463
> > >>>>> -4.171062657
> > >>>>> -5.738499071
> > >>>>> -3.343947176
> > >>>>> -1.944879508
> > >>>>> -5.464109272
> > >>>>> -3.189183392
> > >>>>> -3.684700232
> > >>>>> -2.168303451
> > >>>>> -2.329837082
> > >>>>> -0.761979236
> > >>>>> -2.189025304
> > >>>>> 1.094238807
> > >>>>> -4.812300745
> > >>>>> 0.784198777
> > >>>>> -1.567075922
> > >>>>> 0.143963653
> > >>>>> 1.131119051
> > >>>>> 2.899746353
> > >>>>> -0.498719993
> > >>>>> 3.121623505 I created a time series object with 24 annual
> > >>>>> observations. I
> > >>>>> didn't include dates because there isn't an observation for
> > every
> > >>>>> year.
> > >>>>>
> > >>>>> tsdata<-ts(read.csv("...\\Pre2001LaunchDateTraining.csv"),
> > start =
> > >>>>> c(1,1),
> > >>>>> end = c(24,1), frequency = 1) I then created a time series
> > object
> > >>>>> using
> > >>>>> the
> > >>>>> Arima() function. fitdata <-
> > Arima(tsdata,c(2,0,0),include.drift =
> > >>>>> "true")
> > >>>>> After executing I get this error: Error in (order[2] +
> > >>>>> seasonal$order[2]) >
> > >>>>> 1 & include.drift: operations are possible only for numeric,
> > >>>>> logical or
> > >>>>> complex types Traceback: 1. Arima(tsdata, c(2, 0, 0),
> > include.drift =
> > >>>>> "true")
> > >>>>> Any help would be greatly appreciated!
> > >>>>>
> > >>>>> [[alternative HTML version deleted]]
> > >>>>>
> > >>>>> ______________________________________________
> > >>>>> R-help using r-project.org <mailto:R-help using r-project.org> mailing
> > list -- To UNSUBSCRIBE and more, see
> > >>>>> 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.
> > >>>>>
> > >>>>
> > >>
> > >> [[alternative HTML version deleted]]
> > >>
> > >> ______________________________________________
> > >> R-help using r-project.org <mailto:R-help using r-project.org> mailing list
> > -- To UNSUBSCRIBE and more, see
> > >> 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.
> > >>
> > >
> > > ______________________________________________
> > > R-help using r-project.org <mailto:R-help using r-project.org> mailing list
> > -- To UNSUBSCRIBE and more, see
> > > 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.
> >
>
[[alternative HTML version deleted]]
More information about the R-help
mailing list