answer to question on NHSR slack

nhsr-interval-counter.R

library(tidyverse)
library(lubridate)

## each id is a person, and each row is one episode
## the two episode for id 1 overlap!
example_dates <- tibble(
  id = c(1,1,2,2),
  date_start = ymd(c("2020-01-01", "2020-01-03","2020-04-01", "2020-04-15")),
  date_end = ymd(c("2020-01-05", "2020-01-10", "2020-04-04", "2020-04-16"))
) #%>%
#mutate(episode = interval(date_start, date_end))

library(data.table)

DT <- setDT(copy(example_dates))


counter <- function(DT, start, stop, id) {
  DT[, .(start = min(start), 
         stop = max(end)),
     by=.(id, group=cumsum(c(1, tail(start, -1) > head(end, -1))))]
  
}

counter(DT, date_start, date_end, id)


# id group      start       stop
# 1:  1     1 2020-01-01 2020-01-10
# 2:  2     2 2020-04-01 2020-04-04
# 3:  2     3 2020-04-15 2020-04-16

# tim taylor data 

dat <- data.frame(
  id = c(1,1,2,2,1),
  date_start = ymd(c("2020-01-01", "2020-01-03","2020-04-01", "2020-04-15", "2020-05-01")),
  date_end = ymd(c("2020-01-05", "2020-01-10", "2020-04-04", "2020-04-16", "2020-10-01"))
)

dat <- replicate(125000, dat, simplify = FALSE)
dat <- .mapply(
  FUN = function(x, y) {
    x$id <- x$id + y
    x
  },
  dots = list(x = dat, y = seq_len(125000)),
  MoreArgs = NULL
)
dat <- rbindlist(dat)

res <- counter(dat, date_start, date_end, id)

res[id==2]
# id group      start       stop
# 1:  2     1 2020-01-01 2020-01-10
# 2:  2     4 2020-05-01 2020-10-01