Splitting a dataframe into several dataframes

Question

I have a dataframe that I need split into several dataframes, based on regex searches. There is no set pattern to the searches, ie sometimes there is a single regex, sometime a combination of several. Here is a minimal example with just one set of rows extracted:

Name <- c("John", "Jane", "Arthur", "Maggie")
Age <- c(20, 30, 31, 33)
City <- c("London", "Paris", "New York", "Delhi")

main_df <- data.frame(Name, Age, City)

sub_df <- main_df %>% 
  filter(grepl("J", Name))

main_df <- main_df %>% 
  filter(!grepl("J", Name))

Note that I am extracting some rows into a new dataframe, then deleting the extracted rows from the main dataframe.

I am looking for a single line command to do this. Help appreciated, especially if using dplyr .

Answer 1

We can write a function like

split_df <- function(df, char) {
  split(df, grepl(char, df$Name))
}

new_df <- split_df(main_df, "J")

new_df[[1]]
#    Name Age     City
#3 Arthur  31 New York
#4 Maggie  33    Delhi

new_df[[2]]
#  Name Age   City
#1 John  20 London
#2 Jane  30  Paris

In place of char make sure to pass appropriate character to split on. You can also use regex for char like ^J (starts with J) or J$ (ends with J) etc.

For example,

new_df <- split_df(main_df, "^J")

would give same output as above.

Answer 2

I think the following will allow you to extract rows based on multiple conditions from the original df and delete them from the original, using dplyr as requested.

Name <- c("John", "Jane", "Arthur", "Maggie")
Age <- c(20, 30, 31, 33)
City <- c("London", "Paris", "New York", "Delhi")
main_df <- data.frame(Name, Age, City, stringsAsFactors = F)
conditions <- c(grepl("J",main_df$Name)) # works with several conditions as well
extractanddelete <- function(x, conditions) {
  condf <- data.frame(conditions)
  #fullcondition <- sapply(conditions, all)
  newdfs.list <- lapply(1:ncol(condf), function(i) x %>% filter(condf[,i]))
  newmain <<- x
  notcondf <- !condf
  sapply(1:ncol(condf), function(i) newmain <<- newmain %>% filter(notcondf[,i]))
  return(newdfs.list)
}
ndflist <- extractanddelete(main_df, conditions)
newmain
ndflist
> newmain
    Name Age     City
1 Arthur  31 New York
2 Maggie  33    Delhi
> ndflist
[[1]]
  Name Age   City
1 John  20 London
2 Jane  30  Paris

You receive a list containing as many elements as the conditions you use for filtering and deleting.

For completeness, you can then do main_df <- newmain

This solution also works for other conditions than just grepl

Answer 3

I achieve it with mapply() function which apply function assign() to multiple list(vector) arguments.

Note: pos = 1 is necessary

mapply(FUN = assign, x = c("main_df", "sub_df"),
                     value = split(main_df, grepl("J", main_df$Name)),
                     pos = 1)

main_df

#     Name Age     City
# 3 Arthur  31 New York
# 4 Maggie  33    Delhi

sub_df

#   Name Age   City
# 1 John  20 London
# 2 Jane  30  Paris