Tidyeval in a non-quasiquotation context (psych::describeBy(group))

Question

I am trying to make a function that will return a gt table of the results of the common psych::describe() and psych::describeBy() functions.

the problem is that describeBy() expects the group argument to be of the form data$vector rather than just group = vector.

So

describeBy(data, data$group) 
# works

describeBy(data, group)
# doesn't work

this is fine for general use--just a thing to remember. But to use it in a shiny app, I really need it to take the form of the second call in the above.

here is my function so far:

description<-function(data, group = NULL, fast = TRUE, ...) {
  grp<-paste0(deparse(substitute(group)))
  #print(grp)
  
  
    if(is.null(group)) {
      data %>%
        psych::describe(fast = fast, ...) %>%
        tibble::rownames_to_column() %>%
        dplyr::select(-c(vars)) %>%
        dplyr::mutate(dplyr::across(is.numeric, round, 2)) %>%
        gt::gt() %>%
        gt::tab_options(
          column_labels.font.size = "small",
          table.font.size = "small",
          row_group.font.size = "small",
          data_row.padding = px(3)
        ) %>% 
        tab_header(
          title = paste0("Data Description") 
        )
    } else {
      data %>%
        psych::describeBy(group = group, fast = fast, mat= TRUE, ...) %>%
        tibble::rownames_to_column() %>%
        select(-c(item, vars)) %>%
        dplyr::mutate(dplyr::across(is.numeric, round, 2)) %>%
        arrange(group1) %>%
        group_by(group1) %>%
        gt() %>%
        gt::tab_options(
          column_labels.font.size = "small",
          table.font.size = "small",
          row_group.font.size = "small",
          data_row.padding = px(3))
    } %>% 
    tab_header(
      title = paste0("Data Description") ,
      subtitle = paste0("Grouped by: ",  grp )
    )
}

I've tried a lot of variations on enquo, names(), !!, etc. in the group = call, but I can't quite get it to work right. Can anyone help?

Answer 1

psych::describeBy does not contain a data argument. Hence you cannot use group directly from the data. If you need to use that, you would better use the formula option. eg:

psych::describeBy(iris~Species)

if at all you are writing a function, then do:

my_describe <- function(data, group = NULL, fast = TRUE){
  if(deparse(substitute(group)) == 'NULL')
    psych::describe(data)
  else  psych::describeBy(as.formula(substitute(data~group)))
}

Note: Note the best way to go about it but it works