How to maintain factors when using plot_ly inside a function in R

Question

I am trying to put a plot_ly barplot inside a function so that I can change the variable to plot when needed. The code works perfectly outside of the function, but does not work in the context of being inside a function. The problem is that plot_ly barplot no longer treats the x-variable as a factor, despite it being a factor in the data.frame (verified after testing).

Also, an explanation as to why I have to use enquo() and {{}} to parse the variables in the dplyr context and the plotly context, respectively would be appreciated.

Code (comments are attempts at restoring the factor):

library(plotly)
library(dplyr)
n <- 100
df <- data.frame(x = as.factor(sample(1:3, n, replace = T)), 
                   y = rnorm(n))

summary_plot <- function(df, var){
  quo_var <- enquo(var)
  
  data.frame(df %>%
               group_by({{var}}) %>%
               summarise(n = n())) %>%
  plot_ly(
    # x = ~as.factor(quo_var),
    # x = as.factor(~quo_var),
    x = ~quo_var,
    y = ~n,
    type = "bar"
  )
  
}

summary_plot(df, x)

Answer 1

I couldn't reproduce your error. It works as factors for me when using either numbers or letters too. Maybe try updating your plotly::, ggplot:: and dplyr:: packages?

• The enquo() is required to quote the variable held within the function argument, otherwise the group_by and plotly will look for a column in your dataset called var and fail. The normal way to pass this to the inside functions would be !!quo_var but since plot_ly() uses the formula notation to look for variables within the data frame with x = ~.. you didn't need the !!, and you can also use x = quo_var without the ~
• Now, group_by() which is a part of tidyverse has a shortcut to avoid the enquo() .. !!. workflow by directly using {{var}} as a convenient shorthand

When you have the data-variable in a function argument (i.e. an env-variable that holds a promise2), you need to embrace the argument by surrounding it in doubled braces, like filter(df, {{ var }}). programming with dplyr/Indirection section