How to find consecutive values <2 with length >=3 in R?

Question

I have some data and I want to find consecutive values <=2, and the length of the consecutive values should >3. My objectives are (1) find the length of different groups of consecutive values and (2) find the location of the first value in each group. I have tried the following code:

set.seed(100)
pre = sample(x=1:5, size = 90, replace = T)
which(pre<=2)

and this will produce the result below:

[1]  1  2  4  8 10 13 14 17 18 19 26 30 33 37 40 41 49 50 51 52 53 54 56 57 58 60 66 69 72 80 85 88 89

So, the conservative values groups include: (1)17, 18, 19; (2)40, 41; (3)49, 59, 51, 52, 53, 54; (4)56, 57, 58; (5)88, 89.

However, as I only need consecutive values with length >=3, group (2) and (5) should be excluded from the results. I wonder how I can do this in R? Thanks for any help.

Answer 1

Use rle to get the lengths of each repeated section, then get the position by adding those lengths, and subset to get the desired runs. In the first line, I have to unclass rle first as otherwise data.frame doesn't know how to handle it.

out <- data.frame(unclass(rle(pre<=2)))
out$pos <- head(cumsum(c(1, out$lengths)), -1)
out[out$lengths>=3  & out$values,c("pos", "lengths")]
##  pos lengths
##   17       3
##   49       6
##   56       3

If you prefer chaining with dplyr, here's a version with that idiom.

rle(pre <= 2) %>% unclass() %>% data.frame() %>%
  mutate(pos = c(1, lengths) %>% cumsum %>% head(-1)) %>%
  filter(lengths >=3 & values) %>% select(pos, lengths)

(In a previous version, I used do.call in the first line, which simply puts the results from rle into a data.frame; do.call simply calls the function specified by its first argument with the second argument as its parameters. It's helpful when you have a list of things (such as rle returns) which you want to use as parameters to a function. The code could certainly be written without that step, it just made it easier to keep the parts together and output only the rows you want.)