[R] question about user written function (newb question)

Gabor Grothendieck ggrothendieck at gmail.com
Sun Feb 26 17:03:32 CET 2012


On Sun, Feb 26, 2012 at 9:06 AM, knavero <knavero at gmail.com> wrote:
> Quick newb question about R relating to the line of code below:
>
> rawCool = read.zoo("cooling.txt", FUN = as.chron, format = "%m/%d/%Y %H:%M",
> sep = "\t", aggregate = function(x) tail(x, 1))
>
> I'm wondering what the specifics are for the argument where it has
> "aggregate = function(x) tail(x, 1)". I understand that it removes the last
> row of duplicates/aggregates in the zoo series. I'm confused as to why
> "tail(x, 1)", a built in function in the utils package, requires the coder
> to treat it as a user written function thus defining the assignment, in this
> case an argument, with "function(x)". Why can't the coder just write
> "tail(x, 1)" instead? Also, with the argument 'x', within tail, I'm assuming
> it's looking at all columns simultaneously within the zoo series? Is that
> correct to say? Thanks.

1.  tail(x, 1) is not a function.  Its the operation of executing the
function tail with the values x and 1.  Even if x existed it would be
an error since aggregate= requires a function, not a value.   One
could have written aggregate = tail except that the default is 6
elements whereas we want 1.

One could write this where tail1 is defined to be the same as tail
except its hard coded to use the last element of x only:

tail1 <- function(x) tail(x, 1)
read.zoo(...whatever..., aggregate = tail1)

2.  aggregate= ultimately calls aggregate in the core of R and that
works by looking at the columns one by one.   For a particular column
it splits it into groups (in the case of aggregate.zoo the groups are
defined by rows having the same times) and then for each group
applying the indicated function.

See ?read.zoo and the vignette("zoo-read") for more.

-- 
Statistics & Software Consulting
GKX Group, GKX Associates Inc.
tel: 1-877-GKX-GROUP
email: ggrothendieck at gmail.com



More information about the R-help mailing list