adjust the elements of a column to get a cumsum equal to zero

Question

I have this columns in a bigger dataset (here i just report asset "x" but there are different, hence the idea is to replicate the process for every asset):

df <- structure(list(
        asset = c("x", "x", "x", "x", "x", "x", "x", "x", "x", "x", "x")
        col1 =  c(10, 10, -22, 11, -13, 15, -7, -10, 10, -5, 3),  
        cumsum(col1) = c(10, 20, -2, 9, -4, 11, 4, -6, 4, -1, 2), 
        class = "data.frame", row.names = c(NA, -11L)
     )

I want to correct the negative number in col1 such that the cumsum(col1) becomes equal to

cumsum(col1) = c(10, 20, 0, 11, 0, 15, 8, 0, 10, 5, 8)

To get that result I need to correct the col1 number iff the negative number is bigger than the cumsum of the previous number. For example the -22 in third position should become -20 to match the cumsum of the previous 10+10 Then the -13 should become equal to -11 and the -10 should become -8, while the last three numbers shouldn't change since they do not cumsum to a negative outcome.

So at the end of the correction process I should get

col1 = c(10, 10, -20, 11, -11, 15, -7, -8, 10, -5, 3)
cumsum(col1) = c(10, 20, 0, 11, 0 ,15, 8, 0, 10, 5, 8)

In the process of correction I think that the mechanism should be (I don't know how to do it with R, but I get something in theoretical terms) :

• group_by = each group in col1 should be defined by each col1(row) greater than the cumsum of its previous rows and restard whenever the col1(row) is greater than the previous elements cumsum

• iff col1(row) is greater than the previous cumsum, correct the col1(row) with the group cumsum number with a negative sign in front

• cumsum col1 and check again iff the result matches the desired output, hence there should be no negative cumsum values. The min should be equal to 0

in the original dataset I have multiple asset types, hence not only "x" but also "y", "z", and others. Furthermore I need to group_by investors since the same situation can be applied to 4k investors. hence the real dataset is something like this:

df <- structure(list(
        investor = c("1", "1", "1", "2", "2", "2", "3", "3", "4", "4", "4"),
        asset = c("x", "x", "x", "x", "x", "x", "y", "y", "y", "y", "z")
        col1 =  c(10, 10, -22, 11, -13, 15, 9, -10, 10, -5, 3),  
        cumsum(col1) = c(10, 20, -2, 11, -2, 13, 9, -1, 10, 5, 3), 
        class = "data.frame", row.names = c(NA, -11L)
     )

where i need it to become (the code should just take care of group_by(investor, asset))

df <- structure(list(
        investor = c("1", "1", "1", "2", "2", "2", "3", "3", "4", "4", "4"),
        asset = c("x", "x", "x", "x", "x", "x", "y", "y", "y", "y", "z")
        col1 =  c(10, 10, -20, 11, -11, 15, 9, -9, 10, -5, 3),  
        cumsum(col1) = c(10, 20, 0, 11, 0, 15, 9, 0, 10, 5, 3), 
        class = "data.frame", row.names = c(NA, -11L)
     )

I wrote thinking about a dplyr solution since I'm more confortable with that but I don't know if it is possibile to do in dplyr.

Thanks for the help!

Answer 1

We may do this with accumulate

library(dplyr)
library(purrr)
df %>% 
   group_by(asset) %>%
   mutate(col2csum = accumulate(col1,  ~ if(abs(.x + .y) < abs(.y)) 0 else 
       .x + .y)) %>% 
   ungroup

-output

# A tibble: 11 × 3
   asset  col1 col2csum
   <chr> <dbl>    <dbl>
 1 x        10       10
 2 x        10       20
 3 x       -22        0
 4 x        11       11
 5 x       -13        0
 6 x        15       15
 7 x        -7        8
 8 x       -10        0
 9 x        10       10
10 x        -5        5
11 x         3        8

Update

If we want to change the 'col1'

df %>% 
   group_by(asset) %>%
   mutate(col2csum = accumulate(col1,  ~ if(abs(.x + .y) < abs(.y)) 0 else 
       .x + .y), col1 = c(first(col2csum), diff(col2csum))) %>% ungroup

-output

# A tibble: 11 × 3
   asset  col1 col2csum
   <chr> <dbl>    <dbl>
 1 x        10       10
 2 x        10       20
 3 x       -20        0
 4 x        11       11
 5 x       -11        0
 6 x        15       15
 7 x        -7        8
 8 x        -8        0
 9 x        10       10
10 x        -5        5
11 x         3        8

data

df <- structure(list(asset = c("x", "x", "x", "x", "x", "x", "x", "x", 
"x", "x", "x"), col1 = c(10, 10, -22, 11, -13, 15, -7, -10, 10, 
-5, 3)), class = "data.frame", row.names = c(NA, -11L))