calculating new columns with NA in starting columns-CodePudding

I have data.frame(col1 = 1:4, col2 = c(NA,1,2,3), col3 = 5:8)

and want to create extra columns which is based on computations involving the other columns with NA in it, but keep getting NA as a result. I do not want to see NA, but want to see 1, based on df %>% mutate( new = .[[1]] .[[2]])

Any suggestions how to work through this? I would prefer to stay using tidyverse and dplyr

CodePudding user response：

Here's a possible framework: replace NA with 0s, do your calculations, and then use rows_update to replace the original NAs.

library(dplyr)
df %>% 
  replace(is.na(.), 0) %>% 
  mutate(new = .[[1]] - .[[2]]) %>% 
  rows_update(df)

  col1 col2 col3 new
1    1   NA    5   1
2    2    1    6   1
3    3    2    7   1
4    4    3    8   1