Seasonal arima
I realize that some time series data has seasonal
pattern. people thend buy at the end of year. as muslim , we buy a new cloth
before ramadhan beforthe eid fitri. That is why the seller caught the
opportunity to sell more.
The seasonal make people know the pattern of the
data. LIke the data that is will display this article. the data contain
seasonal pattern like wool australia quarterly.
It must different the seasonal with the un
seasonal patten. we can recognize the pattern only we use the long term data.
the pattern of the arima is (p,d,q) p represent
of the autocorelation, di represent of the diffeence and q represent the moving
average. In seasonal we just use onev variable
only, meaning we do not have to use the other variable. Some time seies method
use other variable to
PDQ
P represent of the seasonal autocorrelation
D represent of the differrence
Q represent of the moving average
Therea is the seasonal number such as 12, meaning
of the seasonal in 12 peiod.
Should i diffeence first before i use auto arima?
because the arima will difference my data?
Only long term can caught the seasonal. colect a
year of selling poduct will not caug the seasonal pettern becaus ethe seasonal
just happen at the end of year, Suposedly, at least the reseacher collcet at
least 5 years (perhaps). The five yeas may found the pattern of the data.
to recognize the seoasnonal we use the software
to predict the seasonal. the easy ways to use the auto.arima.the auto arima
diecty suggest the seasonal of the arima.
Running seasonal arima. Lie this steos, I prepae
the woolyrg dataset
use
> arimawool<-auto.arima(woolyrnq)
> summary(arimawool)
Series: woolyrnq
ARIMA(1,0,0)(0,1,1)[4]
Coefficients:
ar1 sma1
0.8077 -0.6669
s.e. 0.0629 0.0944
sigma^2 = 175880:
log likelihood = -858
AIC=1722
AICc=1722.21 BIC=1730.23
Training set error measures:
ME RMSE MAE MPE
MAPE
Training set -21.21324 408.6706 294.5528 -0.8019345
5.380456
MASE ACF1
Training set 0.6452355 -0.005812332
After detemine arima I also un the residuals chek
an d I found there is evidence for white noise.
