Update a column based on a condition and groupby

Question

My data is:

Prod   Vend    Capac  Dema   Price
 p1     v2       2      6      1
 p1     v1       3      6      2
 p1     v3       3      6      2
 p2     v1          


        

Answer 1

Here's what I would do:

library(data.table)
setDT(DT)

DT[order(Price), src := pmin(Capac, pmax(Dema - shift(cumsum(Capac), fill=0), 0)), by=Prod]

we can see it matches:

   Prod Vend Capac Dema Price Source src
1:   p1   v2     2    6     1      2   2
2:   p1   v1     3    6     2      3   3
3:   p1   v3     3    6     2      1   1
4:   p2   v1     1    1     1      1   1
5:   p2   v3     2    1     2      0   0
6:   p2   v2     5    1     2      0   0
7:   p3   v1     5    3     3      3   3
8:   p3   v2     3    3     4      0   0
9:   p3   v3     1    3     5      0   0

The logic, partly in pseudocode:

• shift(cumsum(Capac), fill=0) is capacity from cheaper vendors

• max(demand - capacity from cheaper, 0) is residual demand for the vendor

• min(capacity, residual demand) is how much to source from the vendor

.

---

The dplyr analogue:

DT %>% arrange(Price) %>% group_by(Prod) %>% 
  mutate(src = pmin(Capac, pmax(Dema - lag(cumsum(Capac), default=0), 0)))