ARIMA model with lag in 1 and 25 in R

3

Please, when applying the ARIMA model (p, d, q) in a series, the partial autocorrelation function shows a peak in lag 1 and another in 25 and no other statistically significant. Is there any command in R to handle these two lags only, not including 25 regressive coefficients?

    
asked by anonymous 05.11.2015 / 16:29

1 answer

3

Yes it does.

Considering the series lh of the R:

> lh
Time Series:
Start = 1 
End = 48 
Frequency = 1 
 [1] 2.4 2.4 2.4 2.2 2.1 1.5 2.3 2.3 2.5 2.0 1.9 1.7 2.2 1.8 3.2 3.2 2.7 2.2 2.2 1.9 1.9
[22] 1.8 2.7 3.0 2.3 2.0 2.0 2.9 2.9 2.7 2.7 2.3 2.6 2.4 1.8 1.7 1.5 1.4 2.1 3.3 3.5 3.5
[43] 3.1 2.6 2.1 3.4 3.0 2.9

Set the model like this:

> arima(lh, order = c(1,1,1), fixed = c(NA, 0))

Call:
arima(x = lh, order = c(1, 1, 1), fixed = c(NA, 0))

Coefficients:
          ar1  ma1
      -0.0404    0
s.e.   0.1443    0

sigma^2 estimated as 0.2525:  log likelihood = -34.35,  aic = 72.7

In this case, I'm saying that the parameter AR1 is free (estimated by the model) and MA1 is equal to zero by means of the argument fixed .

In your case, if you wanted to set a arima(25,1,0) with only RA coefficients 1 and 25, you could do this:

> arima(lh, order = c(25,1,0), fixed = c(NA, rep(0,23), NA))

Call:
arima(x = lh, order = c(25, 1, 0), fixed = c(NA, rep(0, 23), NA))

Coefficients:
          ar1  ar2  ar3  ar4  ar5  ar6  ar7  ar8  ar9  ar10  ar11  ar12  ar13  ar14
      -0.0539    0    0    0    0    0    0    0    0     0     0     0     0     0
s.e.   0.1343    0    0    0    0    0    0    0    0     0     0     0     0     0
      ar15  ar16  ar17  ar18  ar19  ar20  ar21  ar22  ar23  ar24    ar25
         0     0     0     0     0     0     0     0     0     0  0.2994
s.e.     0     0     0     0     0     0     0     0     0     0  0.1918

sigma^2 estimated as 0.2297:  log likelihood = -33.3,  aic = 72.6

The argument fixed is always a vector with the number of elements equal to the number of parameters that your model has. You can pre-specify any value for the parameters, but we usually use only 0 (when we do not want that term) and NA (when we want the parameters to be estimated by the model).

    
05.11.2015 / 20:14