Fill missing value based on previous values

Question

I have the following vector:

v <- c(2L, 2L, NA, NA, NA, NA, 8L, NA)

How can I replace missing values with the value of the previous series + 1 so it becomes:

c(2L, 2L, 3L, 3L, 3L, 3L, 8L, 9L)

did you [search](http://stackoverflow.com/questions/2776135/last-observation-carried-forward-in-a-data-frame) or try something? — rawr, Jun 29 '16 at 13:00

score 5 · Accepted Answer · answered Jun 29 '16 at 13:05

5

As mentioned, zoo has a last-observation-carried-forward function. We can add one to it:

library(zoo)
v2 <- na.locf(v) 
v2[is.na(v)] <- v2[is.na(v)] + 1L
#[1] 2 2 3 3 3 3 8 9

answered Jun 29 '16 at 13:05

Pierre L

28,203
6
47
69

2

or in one line: `ifelse(is.na(v), na.locf(v) + 1L, v)` – G. Grothendieck Jun 29 '16 at 14:53
1

Another option may be `cummax(pmax(pmin(v, lag(v)+1, na.rm=TRUE), 0, na.rm=TRUE))` – akrun Jun 29 '16 at 16:37

score 0 · Answer 2 · answered Jun 29 '16 at 13:05

0

Its not pretty or efficient but it gets the answer you want. Hopefully someone else will be able to post some nicer code but this should get you started.

v <- c(2L, 2L, NA, NA, NA, NA, 8L, NA)
last <- NA
vec <- vector()
for (i in 1:length(v)) {
    cur <- v[i]
    if (! is.na(cur) ) {
        val <- cur
        last <- cur
    }
    else  val <- last + 1
    vec[i] <- val
}

vec

answered Jun 29 '16 at 13:05

gowerc

1,039
9
18

I should have been more specific. You are right, it is easy to do with a loop. Was wondering if there was an elegant solution with dplyr or another package. – Philippe Massicotte Jun 29 '16 at 13:06

Fill missing value based on previous values

2 Answers2