How to do calculation based on previous row results in dplyr - r

Question

I performing some calculations where the result of a row is the input to the next.
I'm using a for loop which is quite slow, is there a way I can use dplyr for these types of calculations? example below
df <- data.frame(beginning_on_hand = c(10,0,0,0,0,0,0,0,0,0,0,0),
sales = c(10,9,4,7,3,7,2,6,1,5,7,1),
ship = c(10,9,4,7,3,7,2,6,1,5,7,1))
dfb <- df %>%
mutate(receipts = 0) %>%
mutate(ending_on_hand = 0) %>%
mutate(receipts = lag(ship, 2)) %>%
mutate(receipts = if_else(is.na(receipts), 0, receipts))
> dfb
beginning_on_hand sales ship receipts ending_on_hand
10 10 10 0 0
0 9 9 0 0
0 4 4 10 0
0 7 7 9 0
0 3 3 4 0
0 7 7 7 0
0 2 2 3 0
0 6 6 7 0
0 1 1 2 0
0 5 5 6 0
0 7 7 1 0
0 1 1 5 0
for(i in 1:(nrow(dfb)- 2)) {
dfb$ending_on_hand[i] <- dfb$beginning_on_hand[i] + dfb$receipts[i] - dfb$sales[i]
dfb$beginning_on_hand[i+1] <- dfb$ending_on_hand[i]
}
> dfb
beginning_on_hand sales ship receipts ending_on_hand
1 10 10 10 0 0
2 0 9 9 0 -9
3 -9 4 4 10 -3
4 -3 7 7 9 -1
5 -1 3 3 4 0
6 0 7 7 7 0
7 0 2 2 3 1
8 1 6 6 7 2
9 2 1 1 2 3
10 3 5 5 6 4
11 4 7 7 1 0
12 0 1 1 5 0

Answer 1

I don't have a dplyr solution for this, but here is a data.table solution for this.
df <- data.frame(beginning_on_hand = c(10,0,0,0,0,0,0,0,0,0,0,0),
sales = c(10,9,4,7,3,7,2,6,1,5,7,1),
ship = c(10,9,4,7,3,7,2,6,1,5,7,1))
dfb <- df %>%
mutate(ending_on_hand = 0) %>%
mutate(receipts = lag(ship, 2)) %>%
mutate(receipts = if_else(is.na(receipts), 0, receipts))
dfb<-data.table(dfb)
df.end <- dfb[, ending_on_hand := cumsum(beginning_on_hand + receipts - sales)][,
beginning_on_hand := beginning_on_hand + lag(ending_on_hand, default = 0)]
>df.end
beginning_on_hand sales ship ending_on_hand receipts
1: 10 10 10 0 0
2: 0 9 9 -9 0
3: -9 4 4 -3 10
4: -3 7 7 -1 9
5: -1 3 3 0 4
6: 0 7 7 0 7
7: 0 2 2 1 3
8: 1 6 6 2 7
9: 2 1 1 3 2
10: 3 5 5 4 6
11: 4 7 7 -2 1
12: -2 1 1 2 5
To explain, data.table uses basically lists to comprise the data and displays it in typically a flat-file manner. It uses SQL type instructions to organize and process data. The functions of note used here are cumsum and lag. cumsum calculates all values prior to a particular index, and lag looks for a value above or prior to a given index.