R identifying a row prior to a change in sign

Question

I have a vector:

df <- c(5,9,-8,-7,-1)

How can I identify the position prior to a change in sign? ie df[2]

score 14 · Accepted Answer · answered Apr 05 '12 at 18:14

14

This is pretty simple, if you know about the sign function...

which(diff(sign(df))!=0)
# [1] 2

answered Apr 05 '12 at 18:14

Joshua Ulrich

168,168
29
327
408

1

...and if you know about the `diff` function :) – Tommy Apr 05 '12 at 18:16
Thank you. That's very helpful. – adam.888 Apr 07 '12 at 08:49
Be careful, this answer considers c(0,1) to have a sign change. This may or may not be desired depending on the application! – WetlabStudent Jun 30 '15 at 06:39

score 1 · Answer 2 · answered Apr 05 '12 at 18:16

1

I prefer Joshua's answer, but here's an alternative, more complicated one just for fun:

head(cumsum(rle(sign(df))$lengths),-1)

answered Apr 05 '12 at 18:16

joran

163,977
32
423
453

similarly to the above answer by @JoshuaUlrich, this answer considers c(0,1) to have a sign change. This may or may not be desired depending on the application! – WetlabStudent Jun 30 '15 at 06:41

score -1 · Answer 3 · answered Apr 05 '12 at 18:20

-1

If you want to be a terrible person, you could always use a for loop:

signchange <- function(x) {
    index = 0
    for(i in 1:length(x))
    {
        if(x[i] < 0)
        {
            return (index)
        }
        else
        {
            index = index + 1
        }
    }
    return (index)
}

answered Apr 05 '12 at 18:20

Sandeep Dinesh

1,905
16
19

2

I wasn't the downvote but if you're going to be a terrible person and use a loop you should at least check whether the first element is positive or negative. The function as is detects the first negative value - not the first sign change. – Dason Apr 05 '12 at 20:16

R identifying a row prior to a change in sign

3 Answers3

Linked