CODEWITHSUNDEEP
rtime-seriesautocorrelation
garej
garej

Reputation: 565

How to use "a set of lags (time differences) to retain" in acf() routing in R

There is a couple of lines in R documentation of acf function.

i a set of lags (time differences) to retain.

j a set of series (names or numbers) to retain.

What do they mean and how to use them? I see no relevant example in docs.

(I though that it should be as simple as acf(time_series, i=c(1,2,4,7)) but it throws warning messages and doesn't affect the output.)

example:

time_series = rnorm(100)
acf(time_series, i=c(1,2,4,7))

# There were 12 warnings (...)
# In plot.window(...) : "i" is not a graphical parameter
# ...

Upvotes: 1

Views: 143

Answers (1)

akrun
akrun

Reputation: 887951

There are 3 methods suggested for acf (Extract, plot and print)

methods(class = acf)
#[1] [     plot  print

The S3 method for the Extraction ([) source code returns and it is a starred one

grep("acf", methods("["), value = TRUE)
#[1] "[.acf"

getAnywhere('[.acf')
function (x, i, j) 
{
    if (missing(j)) 
        j <- seq_len(ncol(x$lag))
    ii <- if (missing(i)) 
        seq_len(nrow(x$lag))
    else match(i, x$lag[, 1, 1], nomatch = NA_integer_)
    x$acf <- x$acf[ii, j, j, drop = FALSE]
    x$lag <- x$lag[ii, j, j, drop = FALSE]
    x
}

Thus, the i, and j is based on the Extraction

Upvotes: 1

Related Questions