How to initialize empty data frame (lot of columns at the same time) in R

Question

I found how to initialize an empty data frame with 3 or 4 dimensions. It\'s like

df <- data.frame(Date=as.Date(character()),
             File=character(         


        

Answer 1

Maybe this -

df <- data.frame(matrix(ncol = 10000, nrow = 0))
colnames(df) <- paste0("hello", c(1:10000))

And @joran's suggestion - df <- setNames(data.frame(matrix(ncol = 10000, nrow = 0)),paste0("hello", c(1:10000)))

Answer 2

I would do this using setDF (or setDT, if you prefer data.table as output) and setnames:

library(data.table)

DF <- setnames(setDF(lapply(integer(1e4), function(...) character(0L))),
               paste0("hello", 1:1e4))
head(names(DF))
# [1] "hello1" "hello2" "hello3" "hello4" "hello5" "hello6"

Both steps (setnames and setDF) are more efficient than the base counterparts, since no copies are made.

A benchmark:

library(microbenchmark)

microbenchmark(times = 1000,
               base = {df <- data.frame(matrix(ncol = 10000, nrow = 0))
               colnames(df) <- paste0("hello", c(1:10000))},
               DT = setnames(setDF(lapply(integer(1e4), 
                                          function(...) character(0L))),
                             paste0("hello", 1:1e4)))
# Unit: milliseconds
#  expr      min       lq     mean   median       uq      max neval cld
#  base 26.77218 30.94223 37.30173 36.76721 37.80338 102.2379  1000   b
#    DT 16.68004 23.18865 30.60573 29.18421 36.03590 178.1045  1000  a