Totalize component products from table - r

Question

I need my warehouse to be able to know how many items of each component we need per day. Basically, I have bundled items made of single products, and I want the warehouse to know how many of those single items they should provide in any given date.
I currently have data like this:
date bundle_name totbund prod1 totprod1 prod2 totprod2
06/01/2019 a_bund 1 a 1 b 1
06/01/2019 a 1
06/01/2019 b 2
07/01/2019 b_bund 1 b 2
07/01/2019 b_bund 2 b 4
07/01/2019 b 2
My expected output is this:
date all_item total
06/01/2019 a 2
06/01/2019 b 3
07/01/2019 b 8
Please notice that the bundle_item column can have bundled items or a single item, so it is mixed.

Answer 1

Something like this could work (I use 'a' as an example):
dat = dat %>%
group_by(date) %>%
summarize(a_bund = sum(tot_bund[bundle_name=='a']),
a_prod1 = sum(prod1[totprod1=='a']),
a_prod2 = sum(prod2[totprod2=='a'])) %>%
mutate(a = a_bund+a_prod1+a_prod2)

Answer 2

I wouldn't use the bundled notation, it sounds overly complicated. If you have everything in a row-format, you can use the group_by - summarize functionality of dplyr.
Assuming the data is called 'df'
library(dplyr)
df <- df %>%
select(date, prod = prod1, totprod = totprod1) %>%
filter(prod != "") %>%
bind_rows(df %>% select(date, prod = prod2, totprod = totprod2) %>% filter(prod != "") %>%
group_by(date, prod) %>%
summarize(totprod = sum(totprod))

Answer 3

I commented you needed a better approach at this problem.
I suggest you consider this from a structured database perspective. In such view, your data (and thus your world) is made of tables with differing and complementary information. And when you need to obtain information to solve your problem, you join data from different tables. If you have used excel, then you'll know it as vlookup.
How I'd approach your problem:
Table of components:
First, I'll have a table of components. This would be a very simple table of 3 columns: name of the product, component from which it is made and amount of component needed.
For your example, I'll have
library(data.table)
components <- structure(list(name = c("a", "b", "a_bund", "a_bund", "b_bund"),
component = c("a", "b", "a", "b", "b"),
amount = c(1, 1, 1, 1, 2)),
row.names = c(NA, -5L),
class = c("data.table", "data.frame"))
Which will produce:
components
name component amount
1: a a 1
2: b b 1
3: a_bund a 1
4: a_bund b 1
5: b_bund b 2
Notice that the information contained here is just the same information that you have in your table in columns 4 to 7 (by the way, your table is called "wide", while mine is "long" formatted. Long is much better for machine processing, and it's considered "tidy").
Table of Requests
Now that you have a table for the components, you'll need a table to put how many units of product x your clients need by y date. Do you notice that I separated the information content in both tables? There's one with components and nothing else; and there's one with requests and nothing else. Each thingy in its own basket!
This table I called requests and it's composed of three columns: dates with the date of the request, name with the name of the product requested by the client, and qty with the quantity the client expects of the product. That would be what you have in columns one to three in your data.
requests <- structure(list(dates = structure(c(17902, 17902, 17902, 17903, 17903, 17903), class = "Date"),
name = c("a_bund", "a", "b", "b_bund", "b_bund", "b"),
qty = c(1, 1, 2, 1, 2, 2)),
row.names = c(NA, -6L),
class = c("data.table", "data.frame"))
Which produces:
requests
dates name qty
1: 2019-01-06 a_bund 1
2: 2019-01-06 a 1
3: 2019-01-06 b 2
4: 2019-01-07 b_bund 1
5: 2019-01-07 b_bund 2
6: 2019-01-07 b 2
Joining the tables
With these two tables, you now need to know how many of each component you'll need in any given date. To solve this I'll use the data.table package, please see ?data.table for details.
requests[components, on = "name" ][, sum(qty*amount), by = .(dates, component)]
What is in there?
requests[components, on = "name"] joins the table requests with components by matching elements with the same name. In other words, it brings the component and amount (from components, of course) for each name in requests. Paste the command and see what the result is.
Data.table syntaxis allows "chaining" or passing an intermediate result to a new operation. That's what happens with the ][ sequence: I joined the tables and now am feeding that result into a new operation.
That new operation is sum(qty * amount). It is multiplying (you weren't wrong initially) the number of requested units qty by the amount of each component needed to produce it, and sums it (aggregates it) by = .(dates, component), which seems pretty self-explanatory. (If you come from the excel world, just think about a pivot or dynamic table).
That produces your expected output:
requests[components, on = "name" ][, sum(qty*amount), by = .(dates, component)]
dates component V1
1: 2019-01-06 a 2
2: 2019-01-06 b 3
3: 2019-01-07 b 8
While the result is the same other answers already provided, I hope you see the difference in approaches, and the enhanced usability of this one. If not, just imagine that k_bundle is made of 19 different components ;)