R How to count cumulative time worked by working employees - r

Question

I have a simple table of the following example data. The last cell for employee 9 is intentionally empty to indicate that the employee 9 is still working.
employee ID
group
start_date
end_date
1
systems
12-Jan-20
14-Feb-21
2
biofx
03-Mar-21
07-Sep-22
3
systems
03-Apr-21
06-Jun-22
4
biofx
01-May-21
07-Jun-22
5
systems
01-Oct-21
07-Jun-22
6
biofx
01-Dec-21
01-Sep-22
7
systems
01-Jan-22
01-Oct-22
8
biofx
01-Feb-22
01-Nov-22
9
systems
01-Jun-22
what I would like to calculate and plot is for each day in a range, how many cumulative workdays have been worked by the currently working employees. If I can get that far, I plan to show an area plot colored by group, or something similar. The hope is that this analysis will highlight the dates when the most senior employees left the company.
So far I have imported and lubridated my table:
#loads the table above with NA in the empty cell
DATES = read_excel(fname)
#example date range. Will likely use the minimum start date eventually
date_range = as_date(mdy("1-1-20"):mdy("1-1-23"))
#reformat the date columns and create an interval
DATES = DATES %>%
mutate(start_date_ymd = ymd(start_date)) %>%
mutate(end_date_ymd = ymd(end_date)) %>%
select(-start_date, -end_date) %>%
mutate(work_interval = interval(start_date_ymd, end_date_ymd))
# naive start - can I just plot the number of workers working on each day?
num_workers<- sapply(date_range, function(x) sum(x %within% DATES$work_interval))
tibble(date_range, num_workers) %>%
ggplot(aes(x=date_range, y=num_workers)) +
geom_point()
Although the last couple of lines above aren't quite what I want - Why don't I see data for worker 8 working up to november?
But even when I figure out why my plot is wrong, I'm really needing some direction about how to calculate on each day the sum of the days worked for all employees working that day.

Answer 1

You can expand the list of all days between start and end. Then summarize by date.
# Basic dataset. Fill in missing end date.
df <- read_table("employee_ID group start_date end_date
1 systems 12-Jan-20 14-Feb-21
2 biofx 03-Mar-21 07-Sep-22
3 systems 03-Apr-21 06-Jun-22
4 biofx 01-May-21 07-Jun-22
5 systems 01-Oct-21 07-Jun-22
6 biofx 01-Dec-21 01-Sep-22
7 systems 01-Jan-22 01-Oct-22
8 biofx 01-Feb-22 01-Nov-22
9 systems 01-Jun-22 ") %>%
mutate(across(ends_with("date"), lubridate::dmy)) %>%
replace_na(list(end_date =lubridate::today()))
# Expand by date:
df2 <- df %>%
mutate(days = map2(start_date, end_date, ~seq(1L, as.integer(.y - .x), by = 1L))) %>%
unnest(days) %>%
mutate(date = start_date + lubridate::days(days)) %>%
select(-start_date, -end_date)
# Summarize by date:
df3 <- df2 %>%
group_by(date, group) %>%
summarize(num_workers = n(),
total_experience = sum(days))
# Plot cumulative days worked
df3 %>%
ggplot(aes(date, total_experience, fill = group)) +
geom_col()
You can clearly see the days when people leave, and how much experience they took with them.