Add Max/Min of differences for numeric variables in "Differences" table of comparedf

[sheramin]

I'm using comparedf function to cross check two tables which usually contain calculated variables, such as BMI. Because the two tables were built by two different programmers, they could have used different rounding which in this case we might have some differences but they're not really significant - just rounding and decimal differences. I think adding a min.diff and max.diff columns in the diffs.byvar.table table for numeric variables could be really helpful in such cases to show the range of differences, so if the differences are reasonable we can skip otherwise to check the details. I know we can print the whole table by setting the control option and visually check this, but that's time consuming when you have a ton of variables. Following is a visual example of what I'm trying to say. (The example datasets are from admiral.test::admiral_vs package. BMI has been calculated by reshaping the dataset).

Below I tried to implement a solution in a very basic and simple way:

add_xy_diff <- compareres$diffs.table %>% 
  mutate(
    values.x = unlist(values.x),
    values.y = unlist(values.y),
    xy.diff = case_when(
      is.numeric(values.x) & is.numeric(values.y) ~ values.x - values.y,
      TRUE ~ NA_real_
    )
  )

xy_diff_smry <- add_xy_diff %>% 
  group_by(var.x, var.y) %>% 
  summarise(
    min.diff = min(xy.diff, na.rm = TRUE),
    max.diff = max(xy.diff, na.rm = TRUE)
  )

compareres$diffs.byvar.table %>% 
  left_join(xy_diff_smry, by = c("var.x", "var.y"))