Question

The code:

A <- matrix(NA, nrow = 10, ncol = 3)
colnames(A) <- paste('trial', 1:3)

Let I want to name the third column with something different from trial 3: how may I?

Ok, now:

library(vars)
data(Canada)
var.2c <- VAR(Canada, p = 2, type = "const")
pred <- predict(var.2c, n.ahead = 8, ci = 0.95)

If I want to see the first value of pred, according to what predict method returns here, I have to input

> pred$fcst$e[1]
[1] 962.6557

You can see that I had to specify time series name e to obtain the value I wanted. If the code above was inserted into a function, it would not be possible to call pred$fcst$e[1] without specifying time series name, thereby it would not be possible to pass another time series data frame to our custom function.

Example: let X be a time series data frame. Then

pred.var <- function(X) {
   var.2c <- VAR(X, p = 2, type = 'const')
   pred <- predict(var.2c, n.ahead = 8, ci = .95)
   # return(pred$fcst$...) # Here lies the issue, because function must know X column name
}

As you can see from example above, function pred.var must know the name of the X column whose it must return the prediction but it cannot unless it is read from X.

Any idea?

Thanks,

Answer 1

You can access an element of a list either by $e or [["e"]], where in the second one, you can use any character variable. So these two things give the same value.

pred$fcst$e[1]
pred$fcst[[ colnames(Canada)[1] ]][1]