insightsengineering
rtables - cell numerical (fraction) output alignment
Hi @gmbecker, @anajens
How feasible would be to improve numerical output alignment in a cell? For example
df <- data.frame(
USUBJID = as.character(c(1:15, 1)),
ARM = factor(c("A", "B", "B", "A", "A", "A", "B", "A", "B", "B", "B", "B", "B", "B", "B", "B"),
levels = c("A", "B")),
AETOXGR = factor(c(1, 1, 2, 3, 4, 1, 2, 5, 4, 4, 4, 4, 4, 4, 4, 4), levels = c(1:5)),
stringsAsFactors = FALSE
)
basic_table() %>%
split_cols_by("ARM") %>%
summarize_occurrences_by_grade(var = "AETOXGR") %>%
build_table(df)
A B
-----------------------
1 2 (40%) 1 (9.1%)
2 0 2 (18.2%)
3 1 (20%) 0
4 1 (20%) 8 (72.7%)
5 1 (20%) 0
It think this below formatting would be more nicer (note the alignment to the . for the percentage:
A B
-----------------------
1 2 (40%) 1 ( 9.1%)
2 0 2 (18.2%)
3 1 (20%) 0
4 1 (20%) 8 (72.7%)
5 1 (20%) 0
In other words, the numbers should be aligned according to their significance digits, i.e.
120
22
30
The short answer is ... medium hard. It is on the list of future features (in particular, ones @waddella intends to put in, as most of the formatting stuff has historically been his domain) but its not slotted for immediate work/priority currently.
We agree that it would be nice and will be better one it does that, though
I tried to do it w/o adding new formats and only with rtables. I think the output with left alignment is decent enough. If it is fine I would close this issue @gmbecker @shajoezhu
basic_table() %>%
split_cols_by("ARM") %>%
analyze("AETOXGR", afun = function(x) {
a <- as.list(table(x))
irl <- lapply(a, function(x) {
if (x > 0)
c(x, x / sum(unlist(a)))
else
x
})
irf <- lapply(a, function(x) ifelse(x > 0, "xx (xx.xx%)", "xx"))
in_rows(.list = irl, .formats = irf, .aligns = "left")
}) %>%
build_table(df)
A B
———————————————————————————
1 2 (40.00%) 1 (9.09%)
2 0 2 (18.18%)
3 1 (20.00%) 0
4 1 (20.00%) 8 (72.73%)
5 1 (20.00%) 0
Indeed, I think the choice between left and the newly under development decimal alignments are sufficient to close this issue. Significant digits, which he mentions but are nto actually used in his example, are a separate issue but one we already have an issue for.