How can I extract elements from lists of lists in R?

Question

I have a bunch of lists containing lists within them (generalised linear model output). I want to write a function which will extract several elements from each list and then combine the results into a data frame.

I want to extract modelset[[1]]$likelihood & modelset[[1]]$fixef, modelset[[2]]$likelihood & modelset[[2]]$fixef, etc, and combine the results into a data frame.

Can someone give me an idea of how to do this?

Apologies if my question is confusing: what I am trying to do is beyond my limited programming understanding.

Further information about my list:

modelset: Large list (16 elements, 7.3Mb)
    :List of 29
    ..$ fixef           : Named num [1:2] -1.236 -0.611
    .. ..- attr(*, "names")= chr [1:2] "(Intercept)" "SMIstd"
    ..$ likelihood      :List of 4
    .. ..$ hlik: num 238
    .. ..$ pvh : num 256
    .. ..$ pbvh: num 260
    .. ..$ cAIC: num 567

    ...etc  

Answer 1

In order to solve this elegantly you need to understand that you can use ['…'] instead of $… to access list elements (but you will get a list back instead of an individual element).

So if you want to get the elements likelihood and fixef, you can write:

modelset[[1]][c('likelihood', 'fixef')]

Now you want to do that for each element in modelset. That’s what lapply does:

lapply(modelset, function (x) x[c('likelihood', 'fixef')])

This works, but it’s not very R-like.

You see, in R, almost everything is a function. […] is calling a function named [ (but since [ is a special symbol for R, in needs to be quoted in backticks: `[`). So you can instead write this:

lapply(modelset, function (x) `[`(x, c('likelihood', 'fixef')))

Wow, that’s not very readable at all. However, we can now remove the wrapping anonymous function (x), since inside we’re just calling another function, and move the extra arguments to the last parameter of lapply:

lapply(modelset, `[`, c('likelihood', 'fixef'))

This works and is elegant R code.

---

Let’s step back and re-examine what we did here. In effect, we had an expression which looked like this:

lapply(some_list, function (x) f(x, y))

And this call can instead be written as

lapply(some_list, f, y)

We did exactly that, with somelist = modelset, f = `[` and y = c('likelihood', 'fixef').