CODEWITHSUNDEEP
rdataframetibble
Chris Kiniry
Chris Kiniry

Reputation: 529

Create new tibbles for each element in a vector or column

I have a tibble/dataframe called sections that I want to use to create several new tibbles/dataframes. I want to iterate over each row and create a new tibble for each. The first column provides the name of the new tibble and the 2nd and 3rd columns provide the indexes to use on another tibble called my_text.

sections <- structure(list(sections = c("cash_and_bank_sweep", "money_market_funds_non-sweep", 
                                    "equities"), 
                       begin_row = c(325L, 345L, 357L), 
                       end_row = c(345L, 357L, 384L)), 
                  class = c("tbl_df", "tbl", "data.frame"), 
                  row.names = c(NA, -3L))

> sections
# A tibble: 3 x 3
  sections                     begin_row end_row
  <chr>                            <int>   <int>
1 cash_and_bank_sweep                325     345
2 money_market_funds_non-sweep       345     357
3 equities                           357     384

set.seed(1)
my_text <- tibble(Strings = sample(letters, size = 1000, replace = TRUE)

> head(my_text)
# A tibble: 6 x 1
  Strings
  <chr>  
1 y      
2 d      
3 g      
4 a      
5 b      
6 w

So the first tibble I want to create would be cash_and_bank_sweep. Manually I can create as follows:

cash_and_bank_sweep <- tibble(Strings = my_text$Strings[sections$begin_row[1]:sections$end_row[1]])

> head(cash_and_bank_sweep)
# A tibble: 6 x 1
  Strings
  <chr>  
1 e      
2 n      
3 e      
4 k      
5 k      
6 q

Is there some way to efficiently do this with a loop or other construct?

Upvotes: 2

Views: 698

Answers (2)

Ronak Shah
Ronak Shah

Reputation: 388817

We can create a sequence between begin_row and end_row and get the data in long format and do an inner_join with my_text column after adding a row_number() column. 

library(tidyverse)

sections %>%
  mutate(value = map2(begin_row, end_row, `:`)) %>%
  unnest(value) %>%
  select(-begin_row, -end_row) %>%
  inner_join(my_text %>% mutate(row = row_number()), by = c('value' = 'row'))

# A tibble: 62 x 3
#  sections            value Strings
#   <chr>               <int> <chr>  
# 1 cash_and_bank_sweep   325 e      
# 2 cash_and_bank_sweep   326 n      
# 3 cash_and_bank_sweep   327 e      
# 4 cash_and_bank_sweep   328 k      
# 5 cash_and_bank_sweep   329 k      
# 6 cash_and_bank_sweep   330 q      
# 7 cash_and_bank_sweep   331 a      
# 8 cash_and_bank_sweep   332 z      
# 9 cash_and_bank_sweep   333 m      
#10 cash_and_bank_sweep   334 a      
# … with 52 more rows

This will return a single dataframe with all the required rows in it, if you need separate dataframes add %>% group_split(sections) in the chain after the last step i.e inner_join.

Upvotes: 0

akrun
akrun

Reputation: 886948

We can use pmap to create a list of tibbles and if we need as individual objects in the global environment (not recommended), use list2env

library(purrr)
lst1 <- pmap(sections[-1], ~ tibble(Strings = my_text$Strings[..1:..2]))
names(lst1) <- sections[[1]]

list2env(lst1, .GlobalEnv)

Or another option is map2

lst1 <- map2(sections$begin_row, sections$end_row,
             ~ tibble(Strings = my_text$Strings[.x:.y]))
names(lst1) <- sections[[1]]

In base R, this can be done with Map

lst1 <- Map(function(i, j) data.frame(Strings = my_text$Strings[i:j]), 
            sections$begin_row, sections$end_row)
names(lst1) <- sections[[1]]

Or using a for loop

lst1 <- vector('list', nrow(sections))
names(lst1) <- sections[[1]]
for(i in seq_along(lst1)) {
    lst1[[i]] <- data.frame(Strings = my_text$Strings[sections$begin_row[i]:sections$end_row[i]])
   }

Upvotes: 1

Related Questions