####
Two time series questions:
I. FITTING TRANSFER FUNCTIONS WITH LAGS: Consider the following toy example:
dates <- paste(11:21, "/01/2005", sep="") Dates <- as.Date(dates, "%d/%m/%Y") set.seed(1) DF <- data.frame(date=Dates, y=rnorm(11), x=rnorm(11, 3)) arima(DF$y, c(1,0,0), xreg=lag(DF$x, 1))
ar1 intercept lag(DF$x, 1)
-0.3876 -1.1328 0.4280
s.e. 0.3556 0.6417 0.1945
sigma^2 estimated as 0.3807: log likelihood = -10.38, aic = 28.76
arima(DF$y, c(1,0,0), xreg=lag(DF$x, 2))
ar1 intercept lag(DF$x, 2)
-0.3876 -1.1328 0.4280
s.e. 0.3556 0.6417 0.1945
sigma^2 estimated as 0.3807: log likelihood = -10.38, aic = 28.76****I NAIVELY THOUGHT THAT "lag" WOULD DO SOMETHING HERE. Evidently, it didn't. ****The following seems to work:
> arima(DF$y, c(1,0,0), xreg=c(NA, DF$x[-11]))
ar1 intercept c(NA, DF$x[-11])
-0.3994 0.3837 -0.0215
s.e. 0.3107 0.7366 0.2259
sigma^2 estimated as 0.5356: log likelihood = -11.15, aic = 30.31
> arima(DF$y, c(1,0,0), xreg=c(NA, NA, DF$x[-(10:11)]))
ar1 intercept c(NA, NA, DF$x[-(10:11)])
-0.5703 0.9180 -0.1794
s.e. 0.3374 0.6685 0.2115
sigma^2 estimated as 0.4871: log likelihood = -9.73, aic = 27.46Is there a better way?
II. ASSOCIATING A CALENDAR DATE WITH A 'ts' OBJECT
In the previous example, I'd like to convert x and y into "ts" objects, retaining "Dates". Is there a way to do this? The following did not work:
ts(DF$y, start=Dates[1])
Error in Math.difftime((end - start) * frequency + 1.01) : floor not defined for difftime objects
Thanks,
spencer graves______________________________________________ [email protected] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
