通过将R中的标记行分组来将2列中的行相乘

Question

I have a dataframe DF, which has the following data; around 300000 rows

<DF
A B C
1 2 0
2 5 0
4 5 2
4 7 0
7 8 0
9 7 -2
2 5 0
4 7 0
5 1 2
4 7 0
7 8 0
9 7 -2
2 5 0
4 7 0
5 1 2

I want to perform a mathematical operation on the data set with the following logic

Select all rows till first occurrence of 2 in C(ignoring -2 in the middle)

Compute avg of all these of (A*B) and add it column D (Implying all these rows in column D will have the same value)

Select all rows from first occurrence of 2 till the second occurrence

Compute avg of (A*B) for these rows and add it column D

... Do the same till

Select all rows from last occurrence of 2 till the second last occurrence

Compute avg of (A*B) for these rows and add it column D

The result should look like

<Result
A B C D
1 2 0 6
2 5 0 6
4 5 2 34.16667
4 7 0 34.16667
7 8 0 34.16667
9 7 -2 34.16667
2 5 0 34.16667
4 7 0 34.16667
5 1 2 27.85714
4 7 0 27.85714
7 8 0 27.85714
9 7 -2 27.85714
2 5 0 27.85714
4 7 0 27.85714
5 1 2 NA

How to implement this logic in R? Thanks in advance!

Answer 1

Using dplyr ,

library(dplyr)
df <- df %>% 
        mutate(ind = cumsum(C == 2)) %>% 
        group_by(ind) %>% 
        mutate(D = mean(A*B), D = replace(D, n() == 1, NA))

Answer 2

Here is an option with data.table . Convert the 'data.frame' to 'data.table' ( setDT(DF) ), grouped by the cumulative sum of logical vector ( C==2 ), we get the mean of A * B and multiply with the the values generated by NA^(.N==1) ( .N==1 returns a logical vector of TRUE/FALSE for number of rows that are equal to 1 or not and using NA^ converts this to NA/1) so that all groups that have only one element returns NA and others have the mean(A*B) .

library(data.table)
setDT(DF)[,  D := NA^(.N==1)*mean(A*B) , .(grp = cumsum(C==2))]
DF
#    A B  C        D
# 1: 1 2  0  6.00000
# 2: 2 5  0  6.00000
# 3: 4 5  2 34.16667
# 4: 4 7  0 34.16667
# 5: 7 8  0 34.16667
# 6: 9 7 -2 34.16667
# 7: 2 5  0 34.16667
# 8: 4 7  0 34.16667
# 9: 5 1  2 31.66667
#10: 4 7  0 31.66667
#11: 7 8  0 31.66667
#12: 9 7 -2 31.66667
#13: 2 5  0 31.66667
#14: 4 7  0 31.66667
#15: 5 1  2       NA