[R] Package forecast

Joao Santos jcsantos at student.dei.uc.pt
Wed Oct 24 17:04:43 CEST 2007


Hello All,

I trying to use the function auto.arima(....) from package forecast but I
have a problem.

My steps after I used the function auto.arima(...)

I create the time series like this:

>bbrass = scan("C:/Program Files/R/data PTIN/my_file.dat")

>regts.start = ISOdatetime(2006, 7, 1, hour=0, min=0, sec=0, tz="GMT")	#2006
07 01 00
>regts.end = ISOdatetime(2006, 7, 22, hour=2, min=0, sec=0, tz="GMT")	#2006
07 22 02
>regts.zoo <- zooreg(bbrass, regts.start, regts.end, deltat=3600)

>regts.hour <- zoo(coredata(regts.zoo), as.numeric(time(regts.zoo))/(3600))
>regts.ts <- as.ts(regts.hour)

>regts.ts
>Time Series:
Start = 319920 
End = 321439 
Frequency = 1 
   [1] 61 60 60 59 58 58 58 58 58 61 64 65 65 64 64 64 63 63 62 61 60 60 60
59
   [25] 58 58 58 57 57 57 57 56 57 57 58 59 59 59 60 60 60 61 60 60 60 60 60
59
   ......


My data is seasonal(repeat every weak) and I get values every hour. I'm
using arima function from package stats like this:
 
>bb.fit = arima(regts.ts, order=c(3,0,0), seasonal=list(order=c(0,1,0),
period=168)) #period=24(hours)*7(days)=168(values)
Series: regts.ts 
ARIMA(0,1,0)(0,1,0)[168]                    

sigma^2 estimated as 6.138:  log likelihood = -2783.19
AIC = 5568.37   AICc = 5568.37   BIC = NaN
#I going changing the values of p,d and q and select the one with the small
value of AIC. 


Then I find the package forecast and the function auto.arima, but my problem
is that when I use this function the result is this I don't have the
seasonal part:
>fit<-auto.arima(regts.ts, d = NA, D = NA, max.p = 3, max.q = 3,
+             max.P = 2, max.Q = 2, max.order = 5,
+             start.p=0, start.q=0, start.P=0, start.Q=0,
+                         stationary = FALSE, ic = c("aic","aicc", "bic"), 
+             stepwise=TRUE, trace=TRUE)

    ARIMA(0,1,0) with drift         : 6178.487
    ARIMA(0,1,0) with drift         : 6178.487
    ARIMA(1,1,0) with drift         : 5467.095
    ARIMA(0,1,1) with drift         : 5571.326
    ARIMA(2,1,0) with drift         : 5419.149
    ARIMA(2,1,1) with drift         : 5420.55
    ARIMA(3,1,1) with drift         : 5422.054
    ARIMA(2,1,0)                    : 5416.789
    ARIMA(1,1,0)                    : 5464.839
    ARIMA(3,1,0)                    : 5417.929
    ARIMA(2,1,1)                    : 5418.177
    ARIMA(3,1,1)                    : 5419.681

    Best model: ARIMA(2,1,0) 



Another problem in arima(....) is because I've NA values and I want to plot
the fitted model with the values created for filling the NA's. 
I used na.interp(regts.ts) but the seasonality vanish.    
ex:bb.fit <- arima(na.interp(regts.ts), order=c(3,0,0),
seasonal=list(order=c(0,1,0), period=168, method = "ML"))


Any ideas to solve this problems?


Sorry the big mail but I don't find other way to explain my problem.

Thanks in advance,

João Santos

  
-- 
View this message in context: http://www.nabble.com/Package-forecast-tf4684066.html#a13384789
Sent from the R help mailing list archive at Nabble.com.



More information about the R-help mailing list