CODEWITHSUNDEEP
rdplyrmultiple-columnstidyr
Costin
Costin

Reputation: 3029

separate list column into columns in R

Given a dataframe with 2 columns, id and value, I want to transform it into a dataframe with more columns containing the id and the quantiles from the column value: q0, q25, q50, q75, q100.

I do not know how to separate a column containing a list into more columns containing its values. Of course, all the lists have the same length.

Here is an example: 

library(dplyr)
library(tidyr)

set.seed(0)
df <- data.frame(id = rep(c("Alice", "Bob"), each = 10),
                 value = round(rnorm(20) * 10))
> df

      id value
1  Alice    13
2  Alice    -3
3  Alice    13
4  Alice    13
5  Alice     4
6  Alice   -15
7  Alice    -9
8  Alice    -3
9  Alice     0
10 Alice    24
11   Bob     8
12   Bob    -8
13   Bob   -11
14   Bob    -3
15   Bob    -3
16   Bob    -4
17   Bob     3
18   Bob    -9
19   Bob     4
20   Bob   -12

df_quantiles <- df %>% 
  group_by(id) %>% 
  summarise( quantiles = list(quantile(value))) %>% 
  ungroup()
> df_quantiles

    # A tibble: 2 x 2
          id quantiles
          
    1  Alice 
    2    Bob 

> df_quantiles$quantiles

[[1]]
  0%  25%  50%  75% 100% 
 -15   -3    2   13   24 

[[2]]
    0%    25%    50%    75%   100% 
-12.00  -8.75  -3.50   1.50   8.00

The next command doesn't do the work. Can you please help me with the good separate call? Is there any other method to get the result?

> df_quantiles %>%
+ separate(quantiles, paste0("q", seq(0,5)))
# A tibble: 2 x 7
      id    q0    q1    q2    q3    q4    q5
*       
1  Alice     c    15     3     2    13    24
2    Bob     c    12     8    75     3     5
Warning message:
Too many values at 2 locations: 1, 2

What I expect is this dataframe:

    id      q0%    q25%    q50%    q75%   q100%
1  Alice    -15      -3       2      13      24
2    Bob -12.00   -8.75   -3.50    1.50    8.00

Upvotes: 4

Views: 1687

Answers (4)

Erwan LE PENNEC
Erwan LE PENNEC

Reputation: 526

A combination of list, as_tibble from tibble, as.list and unnest from tidyr does the job

library(tidyverse)
df_quantiles <- df %>% 
  group_by(id) %>% 
  summarise(quantiles = list(as_tibble(as.list(quantile(value))))) %>% unnest() %>%
  ungroup()

Upvotes: 3

Sandipan Dey
Sandipan Dey

Reputation: 23101

How about

cbind.data.frame(id=unique(df$id), do.call(rbind, df_quantiles$quantiles))

with output

id  0%   25%  50%  75% 100%
1 Alice -15 -3.00  2.0 13.0   24
2   Bob -12 -8.75 -3.5  1.5    8

Upvotes: 5

Ronak Shah
Ronak Shah

Reputation: 388807

If you need dplyr solution, you can use it like :

library(dplyr)
df %>% 
  group_by(id) %>%
  do(data.frame(t(quantile(.$value))))

#     id   X0.  X25.  X50.  X75. X100.
#   <fctr> <dbl> <dbl> <dbl> <dbl> <dbl>
#1  Alice   -15 -3.00   2.0  13.0    24
#2    Bob   -12 -8.75  -3.5   1.5     8

Upvotes: 2

akrun
akrun

Reputation: 886938

We can use data.table

library(data.table)
setDT(df)[, as.list(quantile(value)) , by =  id]
#      id  0%   25%  50%  75% 100%
#1: Alice -15 -3.00  2.0 13.0   24
#2:   Bob -12 -8.75 -3.5  1.5    8

Or using dplyr

library(dplyr)
df %>%
   group_by(id) %>% 
   do(data.frame(as.list(quantile(.$value))))
#     id   X0.  X25.  X50.  X75. X100.
#  <fctr> <dbl> <dbl> <dbl> <dbl> <dbl>
#1  Alice   -15 -3.00   2.0  13.0    24
#2    Bob   -12 -8.75  -3.5   1.5     8

Upvotes: 2

Related Questions