R help - Apr 2012 - Using the R predict function to forecast a model fit with auto.arima function

Hello R users,

Hope everyone is doing great. 

I have a dataset that is in .csv format and consists of two columns: one
named Period (which contains dates in the format yyyy_mm) and goes from
1995_10 to 2007_09 and the second column named pcumsdry which is a
volumetric measure and has been formatted as numeric without any commas or
decimals.

I imported the dataset as pauldataset and made use of the auto.arima
function which is great when you want R to suggest an appropriate arima
model to fit to your data.

I successfully got the arima fit by coding:
> paulfit<-auto.arima(pauldataset$pcumsdry)
> paulfit
which produced the following results:
Series: pauldataset$pcumsdry 
ARIMA(2,1,1) with drift         

Coefficients:
         ar1    ar2      ma1       drift
      0.2684  0.109  -0.8906  -15265.776
s.e.  0.1145  0.102   0.0798    8047.169

sigma^2 estimated as 2.869e+11:  log likelihood=-2265.02
AIC=4540.03   AICc=4540.43   BIC=4555.25

After this, I wanted to make some predictions based on the model suggested
by the auto.arima function. I tried to use the following code to get the
predictions:

variableA.pred<-predict(paulfit, n.ahead=36)
*Error in predict.Arima(paulfit, n.ahead = 36) : 
  'xreg' and 'newxreg' have different numbers of columns: 1 !=
0*>

Now, I have understood, and, according to my research the xreg comes into
play if you add external regressors to your arima fit, but since I am just
working with a univariate time series, I really do not understand the reason
for that error message.

Can anybody help me or tell me how could I do to be able to generate the
predictions?

Best regards,

Paul

--
View this message in context:
http://r.789695.n4.nabble.com/Using-the-R-predict-function-to-forecast-a-model-fit-with-auto-arima-function-tp4590335p4590335.html
Sent from the R help mailing list archive at Nabble.com.

Hmmm.... can you provide the result of dput(paulfit)?

Otherwise, I'm not really sure what might have happened:

library(forecast)
example(auto.arima) # Makes an Arima object "fit" available
predict(fit, n.ahead = 36) # Works fine
forecast(fit, 36) # As does this

Michael

On Thu, Apr 26, 2012 at 12:40 PM, PaulJr <paulbernal07 at gmail.com>
wrote:
> Hello R users,
>
> Hope everyone is doing great.
>
> I have a dataset that is in .csv format and consists of two columns: one
> named Period (which contains dates in the format yyyy_mm) and goes from
> 1995_10 to 2007_09 and the second column named pcumsdry which is a
> volumetric measure and has been formatted as numeric without any commas or
> decimals.
>
> I imported the dataset as pauldataset and made use of the auto.arima
> function which is great when you want R to suggest an appropriate arima
> model to fit to your data.
>
> I successfully got the arima fit by coding:
>
>> paulfit<-auto.arima(pauldataset$pcumsdry)
>> paulfit
>
> which produced the following results:
> Series: pauldataset$pcumsdry
> ARIMA(2,1,1) with drift
>
> Coefficients:
> ? ? ? ? ar1 ? ?ar2 ? ? ?ma1 ? ? ? drift
> ? ? ?0.2684 ?0.109 ?-0.8906 ?-15265.776
> s.e. ?0.1145 ?0.102 ? 0.0798 ? ?8047.169
>
> sigma^2 estimated as 2.869e+11: ?log likelihood=-2265.02
> AIC=4540.03 ? AICc=4540.43 ? BIC=4555.25
>
> After this, I wanted to make some predictions based on the model suggested
> by the auto.arima function. I tried to use the following code to get the
> predictions:
>
> variableA.pred<-predict(paulfit, n.ahead=36)
> *Error in predict.Arima(paulfit, n.ahead = 36) :
> ?'xreg' and 'newxreg' have different numbers of columns: 1
!= 0*>
>
> Now, I have understood, and, according to my research the xreg comes into
> play if you add external regressors to your arima fit, but since I am just
> working with a univariate time series, I really do not understand the
reason
> for that error message.
>
> Can anybody help me or tell me how could I do to be able to generate the
> predictions?
>
> Best regards,
>
> Paul
>
> --
> View this message in context:
http://r.789695.n4.nabble.com/Using-the-R-predict-function-to-forecast-a-model-fit-with-auto-arima-function-tp4590335p4590335.html
> Sent from the R help mailing list archive at Nabble.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.