Comparing vectors or factors with NA

Problem

You want to compare two vectors or factors but want comparisons with NA’s to be reported as TRUE or FALSE (instead of NA).

Solution

Suppose you have this data frame with two columns which consist of boolean vectors:

  1. df <- data.frame( a=c(TRUE,TRUE,TRUE,FALSE,FALSE,FALSE,NA,NA,NA),
  2.                   b=c(TRUE,FALSE,NA,TRUE,FALSE,NA,TRUE,FALSE,NA))
  3. df
  4. #>       a     b
  5. #> 1  TRUE  TRUE
  6. #> 2  TRUE FALSE
  7. #> 3  TRUE    NA
  8. #> 4 FALSE  TRUE
  9. #> 5 FALSE FALSE
  10. #> 6 FALSE    NA
  11. #> 7    NA  TRUE
  12. #> 8    NA FALSE
  13. #> 9    NA    NA

Normally, when you compare two vectors or factors containing NA values, the vector of results will have NAs where either of the original items was NA. Depending on your purposes, this may or not be desirable.

  1. df$a == df$b
  2. #> [1]  TRUE FALSE    NA FALSE  TRUE    NA    NA    NA    NA
  4. # The same comparison, but presented as another column in the data frame:
  5. data.frame(df, isSame = (df$a==df$b))
  6. #>       a     b isSame
  7. #> 1  TRUE  TRUE   TRUE
  8. #> 2  TRUE FALSE  FALSE
  9. #> 3  TRUE    NA     NA
  10. #> 4 FALSE  TRUE  FALSE
  11. #> 5 FALSE FALSE   TRUE
  12. #> 6 FALSE    NA     NA
  13. #> 7    NA  TRUE     NA
  14. #> 8    NA FALSE     NA
  15. #> 9    NA    NA     NA

A function for comparing with NA’s

This comparison function will essentially treat NA’s as just another value. If an item in both vectors is NA, then it reports TRUE for that item; if the item is NA in just one vector, it reports FALSE; all other comparisons (between non-NA items) behaves the same.

  1. # This function returns TRUE wherever elements are the same, including NA's,
  2. # and FALSE everywhere else.
  3. compareNA <- function(v1,v2) {
  4.     same <- (v1 == v2) | (is.na(v1) & is.na(v2))
  5.     same[is.na(same)] <- FALSE
  6.     return(same)
  7. }

Examples of the function in use

Comparing boolean vectors:

  1. compareNA(df$a, df$b)
  2. #> [1]  TRUE FALSE FALSE FALSE  TRUE FALSE FALSE FALSE  TRUE
  4. # Same comparison, presented as another column
  5. data.frame(df, isSame = compareNA(df$a,df$b))
  6. #>       a     b isSame
  7. #> 1  TRUE  TRUE   TRUE
  8. #> 2  TRUE FALSE  FALSE
  9. #> 3  TRUE    NA  FALSE
  10. #> 4 FALSE  TRUE  FALSE
  11. #> 5 FALSE FALSE   TRUE
  12. #> 6 FALSE    NA  FALSE
  13. #> 7    NA  TRUE  FALSE
  14. #> 8    NA FALSE  FALSE
  15. #> 9    NA    NA   TRUE

It also works with factors, even if the levels of the factors are in different orders:

  1. # Create sample data frame with factors.
  2. df1 <- data.frame(a = factor(c('x','x','x','y','y','y', NA, NA, NA)),
  3.                   b = factor(c('x','y', NA,'x','y', NA,'x','y', NA)))
  5. # Do the comparison
  6. data.frame(df1, isSame = compareNA(df1$a, df1$b))
  7. #>      a    b isSame
  8. #> 1    x    x   TRUE
  9. #> 2    x    y  FALSE
  10. #> 3    x <NA>  FALSE
  11. #> 4    y    x  FALSE
  12. #> 5    y    y   TRUE
  13. #> 6    y <NA>  FALSE
  14. #> 7 <NA>    x  FALSE
  15. #> 8 <NA>    y  FALSE
  16. #> 9 <NA> <NA>   TRUE
  19. # It still works if the factor levels are arranged in a different order
  20. df1$b <- factor(df1$b, levels=c('y','x'))
  21. data.frame(df1, isSame = compareNA(df1$a, df1$b))
  22. #>      a    b isSame
  23. #> 1    x    x   TRUE
  24. #> 2    x    y  FALSE
  25. #> 3    x <NA>  FALSE
  26. #> 4    y    x  FALSE
  27. #> 5    y    y   TRUE
  28. #> 6    y <NA>  FALSE
  29. #> 7 <NA>    x  FALSE
  30. #> 8 <NA>    y  FALSE
  31. #> 9 <NA> <NA>   TRUE