dplyr / R cumulative sum with reset

I think you can use accumulate() here to help. And i've also made a wrapper function to use for different thresholds

sum_reset_at <- function(thresh) {
  function(x) {
    accumulate(x, ~if_else(.x>=thresh, .y, .x+.y))
  }  
}

tib %>% mutate(c = sum_reset_at(5)(a))
#       t     a     c
#   <dbl> <dbl> <dbl>
# 1     1     2     2
# 2     2     3     5
# 3     3     1     1
# 4     4     2     3
# 5     5     2     5
# 6     6     3     3
tib %>% mutate(c = sum_reset_at(4)(a))
#       t     a     c
#   <dbl> <dbl> <dbl>
# 1     1     2     2
# 2     2     3     5
# 3     3     1     1
# 4     4     2     3
# 5     5     2     5
# 6     6     3     3
tib %>% mutate(c = sum_reset_at(6)(a))
#       t     a     c
#   <dbl> <dbl> <dbl>
# 1     1     2     2
# 2     2     3     5
# 3     3     1     6
# 4     4     2     2
# 5     5     2     4
# 6     6     3     7

if you're interested in the group building based on cumsum < threshold

You can use the following base:: function:

cumSumReset <- function(x, thresh = 4) {
    ans    <- numeric()
    i      <- 0

    while(length(x) > 0) {
        cs_over <- cumsum(x)
        ntimes <- sum( cs_over <= thresh )
        x      <- x[-(1:ntimes)]
        ans <- c(ans, rep(i, ntimes))
        i   <- i + 1
    }
    return(ans)
}

call:

tib %>% mutate(g = cumSumReset(a, 5))

result:

#   A tibble: 6 x 3
#      t     a     g
#  <dbl> <dbl> <dbl>
#1     1     2     0
#2     2     3     0
#3     3     1     1
#4     4     2     1
#5     5     2     1
#6     6     3     2

• with the group g you can now do whatever you like.