I do have a data frame with different categorical and numerical columns with the following schema:

Id | num_col_1 | num_col_2 | num_col_3 | cat_col_1 | cat_col_2

Now I want to draw a combined plot with ggplot where I (box)plot certain numerical columns (num_col_2, num_col_2) with boxplot groups according cat_col_1 factor levels per numerical columns. Along y axis is the spread of the respective selected columns (not other column). So far I couldn' solve this combined task.

Thank you.


1 Answer 1


If I understand right your question, you are looking to plot selected numerical columns against a selected categorical column of your dataset, am I right ?

If so, you can have the use of dplyr, tidyr and ggplot2 packages to achieve this.

Starting with this dataframe:

  id        num1      num2      num3 cat cat2
1  C -0.48892284  1.417909 2.8884577   a    f
2  C -0.62795166  1.472390 1.6625688   c    f
3  B -0.04691673  2.731553 0.9692889   c    e
4  B  0.16261812 -1.152528 2.4308332   a    d
5  C  1.29230591 -1.609465 2.2089074   a    f
6  E -0.46355650 -1.070132 0.4517597   b    f

Basically, you are selecting first your columns of interest (here num1, num2 and cat), then, you reshape data into a longer format using pivot_longer function to obtain something like that:

df %>% select(id, num1, num2, cat) %>%
  pivot_longer(., cols = c(num1,num2), names_to = "Var", values_to = "Val")

# A tibble: 200 x 4
   id    cat   Var       Val
   <fct> <fct> <chr>   <dbl>
 1 C     a     num1  -0.489 
 2 C     a     num2   1.42  
 3 C     c     num1  -0.628 
 4 C     c     num2   1.47  
 5 B     c     num1  -0.0469
 6 B     c     num2   2.73  
 7 B     a     num1   0.163 
 8 B     a     num2  -1.15  
 9 C     a     num1   1.29  
10 C     a     num2  -1.61  
# … with 190 more rows

Finally, you can add to this pipe sequence the plotting part by calling ggplot and geom_boxplot :

df %>% select(id, num1, num2, cat) %>%
  pivot_longer(., cols = c(num1,num2), names_to = "Var", values_to = "Val") %>%
  ggplot(aes(x = Var, y = Val, fill = cat)) +

enter image description here

Is it what you are looking for ?


id <- sample(LETTERS[1:5],100, replace = TRUE)
num1 <- rnorm(100)
num2 <- rnorm(100)*2
num3 <- rnorm(100)+2
cat <- sample(letters[1:3],100, replace = TRUE)
cat2 <- sample(letters[4:6],100, replace = TRUE)
df <- data.frame(id, num1, num2,num3, cat,cat2)
  • $\begingroup$ Yes, this is what I meant, thank you. A bit disappointing that no way to plot this without the intermediary step of pivoting, but the solution works. $\endgroup$
    – Fredrik
    Commented Jan 17, 2020 at 12:53
  • $\begingroup$ Glad that it works for you ;). I agree that the grammar of ggplot could be a little restrictive. But with function like pivot_longer or melt (from data.table package), it's really easy to integrate into a workflow $\endgroup$
    – dc37
    Commented Jan 17, 2020 at 18:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.