Changing the shape of the data into a sequence of 1s and 0s-CodePudding

I am currently working on a data set in which I am trying to track patients using a medication. I have the data in episodes like this

So basically this data says that in the first episode the patient used the medication for 150 days then the patient stopped for 1 day then the patient used the medication for 60 days then stopped for 2 days etc.

What I want to do is to convert this to longitudinal data so I have the: 150 1s then a single 0 then sixty 1s then 2 zeros etc.

How to do that?

CodePudding user response：

data<-data.frame(
  end.episode.gap.days= c(1,2,2, .....),
  episode.duration = c(150,60,30,.....))

values <- unlist(lapply(1:NROW(data), function(i){
  c(rep(1,data$episode.duration[i]),rep(0,data$end.episode.gap.days[i]))} ))

print(values)

You must complete the numbers in .....

CodePudding user response：

This is a slightly modified version of Ricardo's answer that should be able to "carry over" any id columns:


data <- data.frame(
  id = 1:8,
  name = letters[1:8],
  end.episode.gap.days = c(
    1, 2, 2, 5, 14, 62, 29, 0
  ),
  episode.duration = c(
    150,
    60,
    30,
    60,
    210,
    60,
    30,
    15
  )
)


result <- do.call(rbind, lapply(1:nrow(data), function(i) {
  longitudinal_values <- c(rep(1, data$episode.duration[i]), rep(0, data$end.episode.gap.days[i]))
  id_cols <- data[rep(i, length(longitudinal_values)), !(names(data) %in% c("end.episode.gap.days", "episode.duration"))]
  cbind(id_cols, longitudinal_values)
}))