How to recode a range of rows in between two specific values

Question

I have following data frame:

a <- seq(1:14)
b <- c(0, 0, "start", 0, 0, 0, "end", 0, 0, "start", 0, "end", 0, 0)
df <- data.frame(a, b)

 df
a      b
1      0
2      0
3   start
4      0
5      0
6      0
7    end
8      0
9      0
10  start
11     0
12   end
13     0
14     0

Now, what I want to do is to recode the values in b between "start" and "end" so that:

 df
a      b
1      0
2      0
3   start
4      1
5      1
6      1
7    end
8      0
9      0
10  start
11     1
12   end
13     0
14     0

So far, I haven't got any working code. I tried something with which() and between() and inrange() from the data.table package, but I couldn't really figure it out. Any ideas how to solve this?

Answer 1

Given

df <- data.frame(a, b, stringsAsFactors = FALSE)
#                      ^^^^^^^^^^^^^^^^^^^^^^^^

We can do

idx <- (cumsum(b == "start") - cumsum(b == "end") - (b == "start")) == 1
df <- transform(df, b = replace(b, idx, "1"))
df
#    a     b
#1   1     0
#2   2     0
#3   3 start
#4   4     1
#5   5     1
#6   6     1
#7   7   end
#8   8     0
#9   9     0
#10 10 start
#11 11     1
#12 12   end
#13 13     0
#14 14     0

---

idx is TRUE for elements between "start" and "end".

When we call cumsum(b == "start") - cumsum(b == "end") we are almost there

cumsum(b == "start") - cumsum(b == "end")
# [1] 0 0 1 1 1 1 0 0 0 1 1 0 0 0

We only need to set the positions to zero where b == "start", i.e.

cumsum(b == "start") - cumsum(b == "end") - b == "start"
# [1] 0 0 0 1 1 1 0 0 0 0 1 0 0 0

Test if this vector is 1 to make it logical

idx <- (cumsum(b == "start") - cumsum(b == "end") - (b == "start")) == 1

Result

idx
[1] FALSE FALSE FALSE  TRUE  TRUE  TRUE FALSE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE

We use this logical vector to replace the respective elements of b with "1".

Answer 2

A more compact answer from @RonakShah comment is,

df$b[unlist(mapply(`:`, which(df$b == "start") + 1, which(df$b == "end") - 1))] <- 1

Original Answer

Similar logic to the above compact answer, using lapply, here we find the start and end positions, map this to a list and find the index, then replace the index with 1,

starting <- which(b == "start")
ending <- which(b == "end")
my.ls <- lapply(Map(c, starting, ending), function(x) (x[1]+1):(x[2]-1))

index <- unlist(my.ls)
b[index] <- 1


df <- data.frame(a, b)
df
a     b
1   1     0
2   2     0
3   3 start
4   4     1
5   5     1
6   6     1
7   7   end
8   8     0
9   9     0
10 10 start
11 11     1
12 12   end
13 13     0
14 14     0

Old loop answer

You are able to use the which functions as follows, first define all the starting and ending points, then loop through and change them to 1...

a <- seq(1:14)
b <- c(0, 0, "start", 0, 0, 0, "end", 0, 0, "start", 0, "end", 0, 0)

starting <- which(b == "start")
ending <- which(b == "end")

for (i in 1:length(starting)){
  index <- (starting[i]+1):(ending[i]-1)
  b[index] <- 1
}
df <- data.frame(a, b)
df