creating a new column containing plotmath expression using `dplyr::case_when`

Question

I would like to create a new column containing plotmath expressions that I later plan to use somewhere else in the analysis pipeline.

Here is a minimal example along with I tried. For example, here I am trying to create a new column called label that will have a different plotmath expression depending on the value of y column.

This doesn't seem to work:

# loading needed libraries
library(tidyverse)

# creating a dataframe
df <- data.frame(x = c(1:10), y = c(rep("a", 5), rep("b", 5))) %>%
  tibble::as_data_frame(x = .)

# adding a new column with plotmath expression
df %>%
  dplyr::mutate(.data = .,
                label = dplyr::case_when(
                  y == "a" ~ paste(list(
                  "'This is'", "~alpha==", 1
                ), sep = ""),
                y == "b" ~ paste(list(
                  "'This is'", "~beta==", 2
                ), sep = "")))
#> Error in mutate_impl(.data, dots): Evaluation error: `y == "a" ~ paste(list("'This is'", "~alpha==", 1), sep = "")`, `y == "b" ~ paste(list("'This is'", "~beta==", 2), sep = "")` must be length 10 or one, not 3.

Created on 2018-06-26 by the reprex package (v0.2.0).

What are you trying to do in your call to `case_when`? This is not obvious to me. — Tim Biegeleisen, Jun 27 '18 at 01:50
@TimBiegeleisen Here I am trying to create a new column called `label` that will have a different plotmath expression depending on the value of the `y` column. — Indrajeet Patil, Jun 27 '18 at 01:53
you need to use the `collapse =` argument for `paste()` instead of `sep` — alexwhan, Jun 27 '18 at 02:09
when you use `sep`, you end up with a vector of the same length as the list. Using `collapse` will mean that instead you'll get a vector of length 1 — alexwhan, Jun 27 '18 at 02:20
@alexwhan That worked!! Can you post your answer and explanation and I will accept it as an answer. Thanks! — Indrajeet Patil, Jun 27 '18 at 02:23

score 2 · Accepted Answer · answered Jun 27 '18 at 03:03

The error message is showing that each case is returning length 3. This is because when you paste() a list using sep, you'll get a vector of the same length as the list, so

paste(list(
              "'This is'", "~alpha==", 1
            ), sep = "")

returns a vector of length 3, not 1 or 10 as required. If instead, you use the collapse argument of paste(), you'll get a vector of length 1. In context:

df %>%
  dplyr::mutate(.data = .,
            label = dplyr::case_when(
              y == "a" ~ paste(list(
                "'This is'", "~alpha==", 1
              ), collapse = ""),
              y == "b" ~ paste(list(
                "'This is'", "~beta==", 2
              ), collapse = "")))
# A tibble: 10 x 3
#       x y     label             
#   <int> <fct> <chr>             
# 1     1 a     'This is'~alpha==1
# 2     2 a     'This is'~alpha==1
# 3     3 a     'This is'~alpha==1
# 4     4 a     'This is'~alpha==1
# 5     5 a     'This is'~alpha==1
# 6     6 b     'This is'~beta==2 
# 7     7 b     'This is'~beta==2 
# 8     8 b     'This is'~beta==2 
# 9     9 b     'This is'~beta==2 
#10    10 b     'This is'~beta==2

creating a new column containing plotmath expression using `dplyr::case_when`

1 Answers1