List of lists of lists into a data frame with purrr

Question

Suppose we have the following list structure:

bla <- list(lda = list(list(auc1 = 0.85, auc2 = 0.56), list(auc1 = 0.65, auc2 = 0.72)),
            j48 = list(list(auc1 = 0.99, auc2 = 0.81), list(auc1 = 0.61, auc2 = 0.85)),
            c50 = list(list(auc1 = 0.92, auc2 = 0.59), list(auc1 = 0.68, auc2 = 0.80)))

The desired output is a data frame structured as:

    auc1 auc2
lda 0.85 0.56
lda 0.65 0.72
j48 0.99 0.81
j48 0.61 0.85
c50 0.92 0.59
c50 0.68 0.80

My attempt is pasted below. I'm able to purrr each inner list separately using the call:

bla[[1]] %>%
    map(., function(x) c(auc1 = x[["auc1"]],
                         auc2 = x[["auc2"]])) %>%
    map_dfr(., as.list)

Any idea is appreciated.

Answer 1

Surprisingly only using bind_rows gives you what you want.

dplyr::bind_rows(bla)

This returns a tibble and tibbles don't have rownames.

---

You can do this in base R using do.call + rbind.

do.call(rbind, unlist(bla, recursive = FALSE))


#     auc1 auc2
#lda1 0.85 0.56
#lda2 0.65 0.72
#j481 0.99 0.81
#j482 0.61 0.85
#c501 0.92 0.59
#c502 0.68 0.8 

Answer 2

We can also use rbindlist

library(data.table)
rbindlist(bla)