Extract consecutive sequences from vector

Question

I have a vector called v with positive and negative values as well as zeros. The question is, how to extract (in a list) all the continuous sequences of positive numbers, that is the sequences of positive numbers separated by zeros.

Here is v:

v <- c(-75.09619, -38.31229,   0,  57.17792,  65.55923, 108.52735, 104.29929,  32.47125,0,   0,   0,   0, -26.65008, -49.48638, -79.60670,-90.55343, -34.60761,   0, 21.48842, 38.83820, 42.28727, 0)

The output must be something like:

[1] 57.17792,  65.55923, 108.52735, 104.29929,  32.47125
[2] 21.48842 , 38.83820, 42.28727

Any ideas?

Answer 1

Here's a short one using split:

split(v[v>0],cumsum(v==0)[v>0])
#$`1`
#[1]  57.17792  65.55923 108.52735 104.29929  32.47125
#
#$`6`
#[1] 21.48842 38.83820 42.28727

This essentially just subsets v to positive numbers, then uses the cumsum(v==0) counter (which will be stable within a group of positive values, and increase when it hits the immediately following 0), to split this group. Voila!

---

Another attempt using rle:

r <- rle(v>0)
start <- cumsum(r$lengths)[r$values] - r$lengths[r$values] + 1
end <- start + r$lengths[r$values] - 1
Map(function(start,end) v[start:end],start,end)

#[[1]]
#[1]  57.17792  65.55923 108.52735 104.29929  32.47125
#
#[[2]]
#[1] 21.48842 38.83820 42.28727

Answer 2

You can try:

indices <- which(v[v>=0]==0)
x <- Map(function(x,y) setdiff(v[v>=0][(y+1):(x-1)],0),
         indices[2:length(indices)],
         indices[1:(length(indices)-1)])
x[vapply(x,length,1L)>0]
#[[1]]
#[1]  57.17792  65.55923 108.52735 104.29929  32.47125
#[[2]]
#[1] 21.48842 38.83820 42.28727

What I did:

• took the 0 position out of the vector formed with the >=0 values of v
• with the Map function, you can extract the part of the vector between two zeros
• the last line's purpose is to strip from the result the sequences without values (if for instance there are sequences of two or more zeroes in the original vector)