Populate column by adding onto row above using lag() in R-CodePudding

I want to populate an existing column with values that continually add onto the row above. This is easy in Excel, but I haven't figured out a good way to automate it in R.

If we had 2 columns in Excel, A and B, we want cell B2 to =B1 A2, and cell B3 would = B2 A3. How can I do this in R?

#example dataframe
df <- data.frame(A = 0:9, B = c(50,0,0,0,0,0,0,0,0,0))
#desired output
desired <- data.frame(A = 0:9, B = c("NA",51,53,56,60,65,71,78,86,95))

I tried using the lag() function, but it didn't give the correct output.

df <- df %>%
mutate(B = B   lag(A))

So I made a for loop that works, but I feel like there's a better solution.

for(i in 2:nrow(df)){
df$B[i] <- df$B[i-1]   df$A[i]
}

Eventually, I want to iterate this function over every n rows of the whole dataframe, essentially so the summation resets every n rows. (any tips on how to do that would be greatly appreciated!)

CodePudding user response：

cumsum() can be used to get the result you need.

df$B <- cumsum(df$B   df$A)
df
   A  B
1  0 50
2  1 51
3  2 53
4  3 56
5  4 60
6  5 65
7  6 71
8  7 78
9  8 86
10 9 95

CodePudding user response：

This might be close to what you need, and uses tidyverse. Specifically, it uses accumulate from purrr.

Say you want to reset to zero every n rows, you can also use group_by ahead of time.

It was not entirely clear how you'd like to handle the first row; here, it will just use the first B value and ignore the first A value, which looked similar to what you had in the post.

n <- 5

library(tidyverse)

df %>%
  group_by(grp = ceiling(row_number() / n)) %>%
  mutate(B = accumulate(A[-1], sum, .init = B[1]))

Output

       A     B   grp
   <int> <dbl> <dbl>
 1     0    50     1
 2     1    51     1
 3     2    53     1
 4     3    56     1
 5     4    60     1
 6     5     0     2
 7     6     6     2
 8     7    13     2
 9     8    21     2
10     9    30     2