Question Details

No question body available.

Tags

r data-manipulation

Answers (3)

February 6, 2026 Score: 3 Rep: 105,922 Quality: Medium Completeness: 50%

With base R, you can try cummax within ave

> transform(df,new = ave(value >= 99, id, FUN = cummax))
   id       date     time cost value new
1   1 2026-01-24 10:52:44  234    34   0
2   1 2026-01-24 10:54:44  236    68   0
3   1 2026-01-24 11:05:41  353    99   1
4   1 2026-01-25 11:52:44 5352    78   1
5   1 2026-01-25 11:58:14  143    99   1
6   2 2026-01-24 10:02:44  124    99   1
7   2 2026-01-24 10:22:44  636    99   1
8   3 2026-01-24 10:52:12   53    75   0
9   3 2026-01-25 10:52:44  436    99   1
10  3 2026-01-25 11:12:23 9473    13   1
February 6, 2026 Score: 2 Rep: 18,811 Quality: Low Completeness: 50%

In dplyr, you can use cumany, which explicitly identifies all cases after the first TRUE. The + turns it from a boolean to a 1/0/integer

df %>% 
  mutate(new = +(cumany(value == 99)), .by = id)

Output:

#    id       date     time cost value new

1   1 2026-01-24 10:52:44  234    34   0


2   1 2026-01-24 10:54:44  236    68   0


3   1 2026-01-24 11:05:41  353    99   1


4   1 2026-01-25 11:52:44 5352    78   1


5   1 2026-01-25 11:58:14  143    99   1


6   2 2026-01-24 10:02:44  124    99   1


7   2 2026-01-24 10:22:44  636    99   1


8   3 2026-01-24 10:52:12   53    75   0


9   3 2026-01-25 10:52:44  436    99   1


10  3 2026-01-25 11:12:23 9473    13   1
February 6, 2026 Score: 1 Rep: 11,996 Quality: Medium Completeness: 60%

You could cummulative sum the condition value == 99 per group and check if it is >= 1 and if so, turn the boolean to integer using +.

dplyr::mutate(X, new = +(cumsum(value == 99) >= 1), .by = id)

or

transform(X, new = ave(value, id, FUN = \(x) +(cumsum(x == 99) >= 1)))

or

library(data.table)
setDT(X)[, new := +(cumsum(value == 99) >= 1), by = id][]

or as @skbaldur pointed out 

setDT(X)[, new := cummax(value == 99), by = id][]

or collapse

collapse::ftransform(X, new = +(collapse::fcumsum(value == 99, g = id) >= 1)

giving

# id          date     time cost value new

1   1 2026-01-24 10:52:44  234    34   0


2   1 2026-01-24 10:54:44  236    68   0


3   1 2026-01-24 11:05:41  353    99   1


4   1 2026-01-25 11:52:44 5352    78   1


5   1 2026-01-25 11:58:14  143    99   1


6   2 2026-01-24 10:02:44  124    99   1


7   2 2026-01-24 10:22:44  636    99   1


8   3 2026-01-24 10:52:12   53    75   0


9   3 2026-01-25 10:52:44  436    99   1


10  3 2026-01-25 11:12:23 9473    13   1

data

X = 1), .by = id),
  dplyrcumany = dplyr::mutate(X, new = +(dplyr::cumany(value == 99)), .by = id),
  avecummax = transform(X,new = ave(value >= 99, id, FUN = cummax)),
  avecumsum = transform(X,new = ave(value, id, FUN = \(x) +(cumsum(x == 99) >= 1))),
  collapse = collapse::ftransform(X, new = +(collapse::fcumsum(value == 99, g = id) >= 1)),
  m = transform(X, new = ave(value == 99, id, FUN = \(x) as.integer(Reduce(|, x, accumulate = TRUE)))),
  iterations = 5000
) |> dplyr::arrange(median)
A tibble: 6 × 13

  expression        min   median itr/sec memalloc gc/sec nitr  ngc totaltime result        memory              time               gc      
1 collapse         23µs   24.8µs    37016.    1.75KB     14.8  4998     2   135.02ms     
2 avecummax    151.9µs    161µs     6105.   16.19KB     12.2  4990    10   817.42ms     
3 avecumsum    156.3µs  167.7µs     5504.   16.19KB     12.1  4989    11   906.36ms     
4 m             181.7µs  192.5µs     5025.   16.19KB     13.1  4987    13   992.44ms     
5 dplyrcumany   1.34ms   1.38ms      716.   20.38KB     10.9  4925    75      6.87s    
6 dplyr_cumsum   1.34ms   1.38ms      715.   20.38KB     11.0  4924    76      6.89s