[SOLVED] How do I figure cumulative distribution within groups (as cume_dist does) but separate tied values in R?

How do I figure cumulative distribution within groups (as cume_dist does) but separate tied values in R?

Here is a data set where YEAR is a grouping variable.

dat <- data.frame(YEAR = c(rep(1999,4),rep(2002,3)), VALUE = c(1,2,3,2,1,2,3))

I would like to add a column that looks at the VALUE column and says "where, within its year, does this value sit?" I'm having trouble phrasing it concisely, but pretty much exactly what cume_dist does except that cume_dist lumps tied values and I want them separated. cume_dist takes 1,2,2,3 and gives back 0.25, 0.75, 0.75, 1.00 and I need the ties separated: 1,2,2,4 should give 0.25, 0.50, 0.75, 1.0.

Here's a line based on cume_dist that works except for the tied values:

dat %>% group_by(YEAR) %>% mutate(cumdist = cume_dist(VALUE))

I tried to deconstruct cume_dist (which is rank in group divided by size of group) and use row_number, which separates ties, for the numerator and divide it by the number of rows in each year. This gives me the correct numerator:

dat %>% group_by(YEAR) %>% mutate(rownumber = row_number(VALUE))

But how do I divide those ranks by the number of values in each year (that is, divide all the ranks in 1999 by 4, and the ranks from 2002 by 3)?

Solution

Is this what you are after?

dat %>%
  mutate(cumdist = seq.int(n())[order(VALUE)] / n(), .by = YEAR)

which gives

  YEAR VALUE   cumdist
1 1999     1 0.2500000
2 1999     2 0.5000000
3 1999     3 1.0000000
4 1999     2 0.7500000
5 2002     1 0.3333333
6 2002     2 0.6666667
7 2002     3 1.0000000