How to combine 2 variables and ignore NAs

Question

I have some data like these

var1   var2
10     NA
101    NA
NA     86
11     NA
NA     11
NA     61

If one variable is NA then the other one is

Answer 1

rowSums with na.rm = TRUE will do this. (This is your suggested solution really...)

Assuming your data are in a data.frame DF and your comment

If one variable is NA then the other one is not, and vice-versa.

is true.

 DF$var3 <- rowSums(DF[, c('var1','var2')], na.rm = TRUE)

Answer 2

Various methods exist. Here's one way:

var3 <- ifelse(!is.na(var1),var1,var2)

Here it is working on your example:

  var1 <- c(10,101,NA,11,NA,NA)
  var2 <- c(NA,NA,86,NA,11,61)

  var3 <- ifelse(!is.na(var1),var1,var2)

 > var3
 [1]  10 101  86  11  11  61

This method is relatively general - it works with non-numeric data for example:

  var1 <- c("AB","WZ",NA,"MN",NA,NA)
  var2 <- c(NA,NA,"QT",NA,"MN","RS")

  var3 <- ifelse(!is.na(var1),var1,var2)

 > var3
 [1] "AB" "WZ" "QT" "MN" "MN" "RS"

The suggestion of replacing NA with 0 and adding wouldn't work in that case.