Question

I created a list containing two data lists (character array region and a list results) of the same length. (I tried to manage the data in data.frame, but it seems to be complicated to add data to a data.frame).

study = list(
    region  = character(),
    results = list()
)

study$region[1] = "Hamburg"
study$results[[1]]  = data.frame(month=c(1:5), maxTemp=c(-12, -1, 3, 10, 23))


study$region[2]    = "Bremen"
study$results[[2]]  = data.frame(month=c(1:5), maxTemp=c(-9, -1, 6, 10, 21))

str(study)

print("Maximum temperature of all study regions:")
max(study$results[[1:2]]$maxTemp)

I want to find out the maximum temperature of all timepoint of all regions. I can address each region after another by using e.g. max(study$results[[1]]$maxTemp, but when I try to address all regions max(study$results[[1:2]]$maxTemp I receive an error:

Error in study$results[[1:2]]$maxTemp :

$ operator is invalid for atomic vectors

Where is my mistake? How can I address fields of a several data.frames that are saved in a list of a list? And what are atomic vectors?

Answer 1

[[ can only return a single element. I thought [[ would have thrown an error because of that, not the error you are seeing, but reading ?"[" tells what R does with a call such as yours and explains the behaviour (from ?"["):

Recursive (list-like) objects:
….

 ‘[[’ can be applied recursively to lists, so that if the single
 index ‘i’ is a vector of length ‘p’, ‘alist[[i]]’ is equivalent to
 ‘alist[[i1]]...[[ip]]’ providing all but the final indexing
 results in a list.

The reason for your error is this:

> study$results[[c(1,2)]]
[1] -12  -1   3  10  23

which indicates that R really did this

> study$results[[1]][[2]]
[1] -12  -1   3  10  23

i.e. return the second component (column) of the first data frame, which is an atomic vector because R drops the empty dimension. $ can not be used on atomic vectors hence the error.

If you want to iterate over the list that is study$results, lapply() or sapply() are your friends:

> lapply(study$results, function(y) max(y[, "maxTemp"], na.rm = TRUE))
[[1]]
[1] 23

[[2]]
[1] 21

> sapply(study$results, function(y) max(y[, "maxTemp"], na.rm = TRUE))
[1] 23 21

If you popped names on the components in $results you’d get them in the output too:

> names(study$results) <- study$region
> lapply(study$results, function(y) max(y[, "maxTemp"], na.rm = TRUE))
$Hamburg
[1] 23

$Bremen
[1] 21

> sapply(study$results, function(y) max(y[, "maxTemp"], na.rm = TRUE))
Hamburg  Bremen 
     23      21

which is easier to use and then you don’t need the $region component if you wish.