Expand values of a column to n rows before and m rows after after the value in data frame

Question

I have a data.frame representing different time series. In one column, I marked interesting time points (Note: There can be multiple interesting time points per Id):

Id	Time	Value	Interesting
1	0	12	0
1	1	14	0
1	2	11	0
1	3	12	1
1	4	13	0
1	5	14	0
1	6	12	0
1	7	12	0
..	..	..	..
78	128	13

Now, I would like to mark also n time points before and m points afterward as an interesting block. So if n = 2 and m = 3 I would expect this:

Id	Time	Value	Interesting	Block
1	0	12	0	0
1	1	14	0	1
1	2	11	0	1
1	3	12	1	1
1	4	13	0	1
1	5	14	0	1
1	6	12	0	1
1	7	12	0	0
..	..	..	..	..
78	128	13	0	0

At the moment, I use a gaussianSmooth() and a threshold:

df %>% mutate(Block = ifelse(gaussianSmooth(Interesting, sigma = 4) > 0.001, 1, 0))

But this is cumbersome works and only works if n = m. Is there a “simpler” solution where I can easily set how many rows before and after should be changed. Solutions preferable in dplyr/tidyverse.

Answer 1

With group_modify (works for multiple Interesting values too). Get the indices you like: here the position when Interesting == 1, and then iteratively replace surrounding values with 1 (max(0, i - n):min(nrow(.x), i + m)).

library(dplyr)
n = 2
m = 3

df %>% 
  group_by(Id) %>% 
  group_modify(~ {
    idx <- which(.x$Interesting == 1)
    for(i in idx){
      .x$Interesting[max(0, i - n):min(nrow(.x), i + m)] <- 1
    }
    .x
  })

# A tibble: 8 × 4
# Groups:   Id [1]
     Id  Time Value Interesting
  <int> <int> <int>       <dbl>
1     1     0    12           0
2     1     1    14           1
3     1     2    11           1
4     1     3    12           1
5     1     4    13           1
6     1     5    14           1
7     1     6    12           1
8     1     7    12           0