R Script for splitting data frame and then saving separate .csv

split-df-save.R

#----------------------------------------------------------------------------------------
# File: 
# Author: Joseph Flanagan, adopted from https://stackoverflow.com/questions/10002021/split-dataframe-into-multiple-output-files-in-r
# email:   [email protected]
# Purpose: Split a dataframe by group, then save each as separate .csv file
#----------------------------------------------------------------------------------------

# new tidyverse solution with `group_walk`
library(dplyr)
library(readr)

iris %>% 
  group_by(Species) %>% 
  group_walk(~ write_csv(.x, paste0(.y$Species, ".csv")))



# Old version
library(tidyverse)

# Make a copy of iris
iris2 <- iris  

# Split by variable
spt2 <- split(iris2, iris2$Species)

# Save
lapply(names(spt2), function(x){
    write_csv(spt2[[x]], paste(x, ".csv", sep = ""))
    })

Sorry, are you sure that write_csv() from the tidyverse has been deprecated? Where did you see that? And I thought that the difference between write_csv() and write_csv2() is more an issue of whether you want a comma as a delimiter (US and others) or a semi-colon (at least some European countries). And that choice mostly due to whether the locale in question uses a point or a comma as a decimal separator.

I'm sorry, I'm mistaken. The "path" argument inside of write_csv is deprecated, I just quickly read the error message R spilled out and did not realize it was the argument and not the entire function.

The "path" argument inside of write_csv is deprecated

Thanks for the update. I just saw that path = is deprecated and we should now use file = . I updated it so just to avoid the issue completely (although my preference is for arguments after the first to be named).

I also updated the original solution it to take advance of group_walk from dplyr. It's now a much cleaner solution that doesn't rely upon lapply() at all.

I prefer to use readr for most cases just for convenience, but obviously there are a range of alternatives and people can always find another package for their own use cases.

thank you!!!