amend calc_freq

[Analyst-In-The-Wild]

compleeeteeeeeeeeeeeeeeeeeeeeee me!

[milanwiedemann]

I know what lori means but I know I will forget, so here more details ...
when monthly sums get calculated, there may be groups with a count of zero.
the current function does not have a row for these cases in the data that gets returned and therefore it wont show in up in figures.
as this functions purpose is mainly to prepare data for figures this is VERY BAD! and needs t obe fixed.

I think I will use the https://tsibble.tidyverts.org/reference/fill_gaps.html function from the the https://tsibble.tidyverts.org/ package will help ...

I know that @ChrisBeeley wrote something for this a while ago that @Analyst-In-The-Wild and I reused in one of our reports, but it felt a bit hacky and this approach may be better?

[milanwiedemann]

wait wait wait, I think I just found the code, it looks like this I believe:

  complete(month_floor_date = seq.Date(min(month_floor_date), max(month_floor_date), 
                   by = "month")) %>% 
  tidyr::replace_na(list(count = 0))

and I have to admit, looking at it right now it doesnt seem too hacky, maybe I just add something like this to our function

[milanwiedemann]

just for reference, this is the function we are talking about

nottshcMethods/R/helper_funs.R

Lines 53 to 104 in 6cff6ed

	#' Calculate frequencies of groups per month
	#'
	#' @param data Data
	#' @param by Grouping variables used in group_by
	#' @param date_var_name String specifying date variable, needs to be class date
	#' @param .drop_latest_month Logical, specifying whether or not to drop the most recent month
	#' @param .calc_year_month_day_vars Logical, specifying whether to calculate separate year, month, and day variable
	#'
	#' @return
	#' @export
	#'
	#' @examples
	calc_monthly_freq <- function(data, by, date_var_name, .drop_latest_month = TRUE, .calc_year_month_day_vars = TRUE) {
	data <- data %>%
	dplyr::mutate(floor_date_m = lubridate::floor_date( {{date_var_name}} ,
	unit = "month"),
	floor_date_m = lubridate::as_date(floor_date_m))
	if (.drop_latest_month) {
	max_date_m <- max(data$floor_date_m)
	data <- data %>%
	dplyr::filter(floor_date_m != max_date_m)
	}
	data <- data %>%
	dplyr::group_by(dplyr::across(c( {{by}}, floor_date_m))) %>%
	dplyr::summarise(n = dplyr::n())
	if (.calc_year_month_day_vars) {
	data <- data %>%
	mutate(year = lubridate::year(floor_date_m),
	month = lubridate::month(floor_date_m,
	label = TRUE,
	abbr = TRUE),
	day = lubridate::wday(floor_date_m,
	label = TRUE,
	abbr = TRUE)) %>%
	dplyr::relocate(floor_date_m) %>%
	dplyr::relocate(year, month, day,
	.after = floor_date_m)
	}
	data %>%
	dplyr::ungroup()
	}

[milanwiedemann]

@Analyst-In-The-Wild do you feel ready to implement this and make a pull request?
if not, how about we do it together?

[Analyst-In-The-Wild]

@milanwiedemann let’s do it together as my only contribution to this was mentioning complete and you suddenly realised it was missing here then made me be your secretary to make the issue 😝

[Lextuga007]

@Analyst-In-The-Wild complete() code doesn't appear in this function yet. Is this something you still have a record of/local branch at all?