| id | time | value1 | indicator | value2 |
|---|---|---|---|---|
| 1 | 2004 | 1 | 1 | 1 |
| 1 | 2005 | 0 | 0 | 1 |
| 1 | 2006 | -1 | 0 | 1 |
| 1 | 2007 | -3 | 0 | 1 |
| 1 | 2008 | 3 | 0 | 1 |
| 2 | 2004 | 1 | 0 | NA |
| 2 | 2005 | 3 | 0 | NA |
| 2 | 2006 | 0 | 0 | NA |
| 2 | 2007 | 0 | 0 | NA |
| 3 | 2004 | -1 | 1 | -1 |
| 3 | 2005 | -3 | 0 | -1 |
| 3 | 2006 | 4 | 0 | -1 |
| 3 | 2007 | 5 | 0 | -1 |
| 4 | 2004 | 4 | 0 | -4 |
| 4 | 2005 | 5 | 0 | -4 |
| 4 | 2006 | 2 | 0 | -4 |
| 4 | 2007 | -4 | 1 | -4 |
Indicator is a binary variable. The desired output is value 2. If indicator is 0, its value 2 takes a value 2 of the year when indicator is 1. It is grouped by id. If an id does not have any 1 for indicator, value2 takes NA's.
library(dplyr)
df %>%
group_by(id) %>%
mutate(value2 = ifelse(indicator == 1, value1 , NA)
) %>%
tidyr::fill(value2, .direction = "updown")
Result:
id time value1 indicator value2
<int> <int> <int> <int> <int>
1 1 2004 1 1 1
2 1 2005 0 0 1
3 1 2006 -1 0 1
4 1 2007 -3 0 1
5 1 2008 3 0 1
6 2 2004 1 0 NA
7 2 2005 3 0 NA
8 2 2006 0 0 NA
9 2 2007 0 0 NA
10 3 2004 -1 1 -1
11 3 2005 -3 0 -1
12 3 2006 4 0 -1
13 3 2007 5 0 -1
14 4 2004 4 0 -4
15 4 2005 5 0 -4
16 4 2006 2 0 -4
17 4 2007 -4 1 -4