Hello all,
I use arima to fit the model with
fit <- arima(y, order = c(1,0,1), xreg = list.indep, include.mean = TRUE)
and would like to use predict() to forecast:
chn.forecast <- rep(0,times=num.record)
chn.forecast[1] <- y[1]
for (j in 2:num.record){
indep <- c(aa=chn.forecast[j-1],
list.indep[j,2:num.indep]) # this is the newxreg in the forecast.
chn.forecast[j] <- predict(fit, newxreg=indep, n.ahead = 1)
}
However, I got the error message as 'xreg' and 'newxreg' have different numbers
of columns".
So I debug into predict.Arima (as shown in (*)).
(*):
debugging in: predict.Arima(fit, newxreg = indep, n.ahead = 1)
debug: {
myNCOL <- function(x) if (is.null(x))
0
else NCOL(x)
rsd <- object$residuals
xr <- object$call$xreg
xreg <- if (!is.null(xr))
eval.parent(xr)
else NULL
ncxreg <- myNCOL(xreg)
if (myNCOL(newxreg) != ncxreg)
stop("'xreg' and 'newxreg' have different numbers of columns")
class(xreg) <- NULL
xtsp <- tsp(rsd)
n <- length(rsd)
arma <- object$arma
coefs <- object$coef
narma <- sum(arma[1L:4L])
if (length(coefs) > narma) {
if (names(coefs)[narma + 1L] == "intercept") {
xreg <- cbind(intercept = rep(1, n), xreg)
newxreg <- cbind(intercept = rep(1, n.ahead), newxreg)
ncxreg <- ncxreg + 1L
}
xm <- if (narma == 0)
drop(as.matrix(newxreg) %*% coefs)
else drop(as.matrix(newxreg) %*% coefs[-(1L:narma)])
}
else xm <- 0
if (arma[2L] > 0L) {
ma <- coefs[arma[1L] + 1L:arma[2L]]
if (any(Mod(polyroot(c(1, ma))) < 1))
warning("MA part of model is not invertible")
}
if (arma[4L] > 0L) {
ma <- coefs[sum(arma[1L:3L]) + 1L:arma[4L]]
if (any(Mod(polyroot(c(1, ma))) < 1))
warning("seasonal MA part of model is not invertible")
}
z <- KalmanForecast(n.ahead, object$model)
pred <- ts(z[[1L]] + xm, start = xtsp[2L] + deltat(rsd),
frequency = xtsp[3L])
if (se.fit) {
se <- ts(sqrt(z[[2L]] * object$sigma2), start = xtsp[2L] +
deltat(rsd), frequency = xtsp[3L])
return(list(pred = pred, se = se))
}
else return(pred)
}
Browse[3]>
debug: myNCOL <- function(x) if (is.null(x)) 0 else NCOL(x)
Browse[3]>
debug: rsd <- object$residuals
Browse[3]>
debug: xr <- object$call$xreg
Browse[3]>
debug: xreg <- if (!is.null(xr)) eval.parent(xr) else NULL
Browse[3]>
debug: eval.parent(xr)
Browse[3]>
debug: ncxreg <- myNCOL(xreg)
Browse[3]>
debug: if (myNCOL(newxreg) != ncxreg) stop("'xreg' and 'newxreg' have different
numbers of columns")
Browse[3]> head(xreg)
aa dummy1 dummy2 bb cc
[1,] 0.015538 0 0 0.941 1.241
[2,] 0.015478 0 0 0.952 1.185
[3,] 0.015607 0 0 0.955 1.422
[4,] 0.015861 0 0 1.038 1.777
[5,] 0.016005 0 0 1.286 2.118
[6,] 0.016180 0 0 1.351 2.084
Browse[3]> newxreg
aa dummy1 dummy2
bb
0.015478 0.000000 0.000000
0.952000
cc
1.185000
Browse[3]> NCOL(xreg)
[1] 5
Browse[3]> NCOL(newxreg)
[1] 1
Browse[3]> NCOL(t(newxreg))
[1] 5
When comparing the column numbers of xreg and newxreg, the function NCOL is
used. It seems to me that they all have the same number of columns
(ncol(xreg)=ncol(newxreg)=5, yellow highlighted part), however,
ncol(newxreg)=1. If I check the transpose ncol(t(newxreg)), it is 5. So I think
about put t(indep) instead of indep as the newxreg, but got error message(**):
(**)
Error in as.matrix(newxreg) %*% coefs :
requires numeric/complex matrix/vector arguments
In addition: Warning message:
In chn.forecast[j] <- predict(fit, newxreg = t(indep), n.ahead = 1) :
number of items to replace is not a multiple of replacement length
How could I fix such a problem?
Thanks very much!
Cheers,
Rebecca
----------------------------------------------------------------------
This message, and any attachments, is for the intended r...{{dropped:5}}
______________________________________________
R-help@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.
