Determine how much each group is contributing to a total amount in R

I have a dataset similar to the following:

df_initial <- data.frame(total = c(1000, 800),
                         group1 = c(500, 100),
                         group2 = c(0, 600),
                         group3 = c(600, 0),
                         group4 = c(100, 200))

The dataset has multiple fields, each representing a different group. Each row has a total amount and amounts associated with each group. I would like to determine how much each group is contributing to the total amount by allocating the amounts associated with each group in a specific order - i.e., I want to allocate the amounts associated with group1 first, group2 second, etc.

In other words, I'm looking to create df_new, which shows that for the total amount of 1000, group1 contributes 500, group2 contributes 0, group3 contributes 500, and group4 contributes 0.

df_new <- data.frame(total = c(1000, 800), 
                     group1_contribution = c(500, 100), 
                     group2_contribution = c(0, 600),
                     group3_contribution = c(500, 0), 
                     group4_contribution = c(0, 100))

I'd greatly appreciate any advice or ideas about how to do this in R.

I think a basic loop would work well here

remaining <- df_initial[[1]]
df_output <- data.frame(total=remaining)
for (i in 2:ncol(df_initial)) {
  contrib <- pmin(df_initial[[i]], remaining)
  df_output[,names(df_initial)[i]] <- contrib
  remaining <- remaining - contrib
}
df_output
#   total group1 group2 group3 group4
# 1  1000    500      0    500      0
# 2   800    100    600      0    100

It loops over the columns keep track of the "remaining balance". It deducts the smaller of the remaining amount or the column value.

(groups <- grep("^group", names(df_initial), value = TRUE))
# [1] "group1" "group2" "group3" "group4"
df_new <- df_initial
df_new[,groups] <- t(
  apply(df_initial[,c("total", groups)], 1, function(z) pmin(z[-1], pmax(0, Reduce(`-`, z[-1], init=z[1], accumulate=TRUE))[-length(z)]))
)
names(df_new)[2:5] <- paste0(names(df_new)[2:5], "_contribution")
df_new
#   total group1_contribution group2_contribution group3_contribution group4_contribution
# 1  1000                 500                   0                 500                   0
# 2   800                 100                 600                   0                 100

Note: the use of apply works here because all referenced columns are numeric; blending different classes (e.g., string) will break this.
This is why I explicitly use groups to indicate all group# columns, and c("total", groups) to indicate the total and all group# columns.

The setNames step is merely to align the output with your df_new, without it the frame still has all of the correct values, and columns are named group1, etc.

library(dplyr)
library(tidyr)

df_initial |>
  pivot_longer(starts_with("group"),
               names_pattern = "(\\d+)") |>
  mutate(remaining = purrr::accumulate(value, `-`, .init = first(total))[-n()],
         value = ifelse(remaining < 0, value + remaining, pmin(value, remaining)),
         value = replace(value, value<0, 0),
         .by = total) |>
  pivot_wider(names_from = name,
              names_glue = "group{name}_contribution",
              id_cols = total)