Fill contingency table based on total variable

Question

I have a list of stores and I have a product (apples). I ran a system of linear equations to get the column \'var\'; this value represents the amount of apples you will

Answer 1

Here's a tidyverse solution. It relies on there being a net zero of each sku.

If that's the case, then we should be able to line up all the donated items (one row for each unit in the negative vars, sorted by sku) with all the received items (one row for each positive var, sorted by sku). Consequently, the first 5 donated apples are matched with the first 5 received apples, and so on.

Then we total up the total for each sku between each donor and recipient pair and spread so each recipient gets a column.

Edit: corrected sign and added complete to match OP solution

library(tidyverse)
output <- bind_cols(

  # Donors, for whom var is negative
  df %>% filter(var < 0) %>% uncount(-var) %>% select(-var) %>%
    arrange(sku) %>% rename(donor = store),

  # Recipients, for whom var is positive
  df %>% filter(var > 0) %>% uncount(var) %>% 
    arrange(sku) %>% rename(recipient = store)) %>%

  # Summarize and spread by column
  count(donor, recipient, sku) %>%
  complete(donor, recipient, sku, fill = list(n = 0)) %>%
  mutate(recipient = paste0("ship_to_", recipient)) %>%
  spread(recipient, n, fill = 0)


> output
# A tibble: 6 x 8
  donor sku   ship_to_a ship_to_b ship_to_c ship_to_d ship_to_e ship_to_f
                                 
1 a     apple         0         0         0         0         0         0
2 b     apple         0         0         0         0         0         0
3 c     apple         1         4         0         0         1         0
4 d     apple         0         0         0         0         1         0
5 e     apple         0         0         0         0         0         0
6 f     apple         0         0         0         0         3         0