Save a data frame with list-columns as csv file

Question

I have the following data frame that looks like this (3 columns as list).

A tibble: 14 x 4
                                                    clinic_name drop_         


        

Answer 1

Here's another option that may be a little simpler.

Depending on the data, comma separated values could get complicated, so I'm using a bar | for separating values in list columns:

library(tidyverse)

starwars %>% 
  rowwise() %>% 
  mutate_if(is.list, ~paste(unlist(.), collapse = '|')) %>% 
  write.csv('df_starwars.csv', row.names = FALSE)

starwars is one of the dplyr sample dataframes.

Answer 2

exploratory::list_to_text() will convert a list column to a character column. The default is sep = ", ", which I recommend changing to something else if writing to a .csv.

devtools::install_github("exploratory-io/exploratory_func")

list_to_text <- function(column, sep = ", "){
  loadNamespace("stringr")
  ret <- sapply(column, function(x) {
    ret <- stringr::str_c(x, collapse = sep)
    if(identical(ret, character(0))){
      # if it's character(0)
      NA
    } else {
      ret
    }
  })
  as.character(ret)
}

https://github.com/exploratory-io/exploratory_func/blob/master/LICENSE.md

Answer 3

Create a tibble containing list columns:

library(tibble)

clinic_name <- c('bobo center', 'yoyo plaza', 'lolo market')
drop_in_hours <- list(c("Monday: 2 pm - 5 pm", "Tuesday: 4 pm - 7 pm")) 
appointment_hours <- list(c("Monday: 1 pm - 2 pm", "Tuesday: 2 pm - 3 pm")) 
services <- list(c("skin graft", "chicken heart replacement"))

tibb <- data_frame(clinic_name, drop_in_hours, appointment_hours, services)

print(tibb)

Write a general-purpose function that converts any list columns to character type:

set_lists_to_chars <- function(x) {
    if(class(x) == 'list') {
    y <- paste(unlist(x[1]), sep='', collapse=', ')
    } else {
    y <- x 
    }
    return(y)
}

Apply function to tibble with list columns:

new_frame <- data.frame(lapply(tibb, set_lists_to_chars), stringsAsFactors = F)

new_frame

Write newly formatted dataframe as csv file:

write.csv(new_frame, file='Desktop/clinics.csv')

This is a csv file with the list columns expanded as regular strings.

Here is an all-encompassing function. Just pass in your tibble and a filename:

tibble_with_lists_to_csv <- function(tibble_object, file_path_name) {
    set_lists_to_chars <- function(x) { 
        if(class(x) == 'list') { y <- paste(unlist(x[1]), sep='', collapse=', ') } else { y <- x  } 
        return(y) }
    new_frame <- data.frame(lapply(tibble_object, set_lists_to_chars), stringsAsFactors = F)
    write.csv(new_frame, file=file_path_name)
}

Usage:

tibble_with_lists_to_csv(tibb, '~/Desktop/tibb.csv')

Answer 4

I had a similar dataframe with list columns that I wanted to save as csv. I figured out this method. As well as how to turn the columns back into lists.

library(tidyverse)

# create a df with a list column
df <- tibble(x=rep(1:5,each=2), y=LETTERS[1:10]) %>%
  group_by(x) %>%
  summarise(z=list(y))

# this throws an error
write_csv(df, "test.csv")

# convert the list column to a string
df2 <- df %>%
  group_by(x) %>% # where x==unique(x)
  mutate(z=paste(z))

# this works
write_csv(df2, "test.csv")

# read the csv
df3 <- read_csv("test.csv")

# reconstruct original df by parsing the strings
# https://stackoverflow.com/questions/1743698/evaluate-expression-given-as-a-string
df4 <- df3 %>%
  group_by(x) %>% 
  mutate(z=list(eval(parse(text=z))))

Answer 5

Is there any specific reason why you would like to save the columns as a list ? Alternatively, you can use unnest and save it in csv. example below

library(tidyverse)
df_list<-data_frame(abc = letters[1:3], lst = list(1:3, 1:3, 1:3))
df_list %>% unnest() %>% write.csv("list.csv")

further, when you read the file you can nest it back

df <- read.csv("list.csv")[ ,2:3]
df %>% nest(lst)