logical query on mixed (character and numeral) variable

Question

I didn't think this would be so difficult but I've struggled for some time... I hope it's something simple I am missing :) I need to be able to logically query a dataframe containing variables that are a mixture of character strings and numerals.

LAMBDA<-data.frame(cbind(list(1,"lequiv"),list("lequiv",0)))
LAMBDA!=1

What is the best way to achieve this, so that the character strings, which clearly do not equal 1, are reported as FALSE, rather than NA? Thanks.

Answer 1

Instead of testing for elements which are equal to the number 1, you could try testing for elements which are equal to the string "1". For example:

> LAMBDA!="1"
        X1   X2
[1,] FALSE TRUE
[2,]  TRUE TRUE

whereas:

> LAMBDA!=1
        X1   X2
[1,] FALSE   NA
[2,]    NA TRUE

In the first case, R is clever enough to coerce the value 1 in your data frame to be a string before testing for equivalence. In the second case it cannot coerce "lequiv" to a number betore testing for equivalence and therefore returns NA (and a warning message).

Answer 2

I think you're looking for factors here.

If you look at str(LAMBDA) you'll see that X1 is a list with two elements, one stored as numeric, one as character.

If you instead create your list as follows:

l1 <- data.frame(X1=c(1,"lequiv"),
                 X2=c("lequiv",2)
                  )

it will look just the same when printed, but the variables are represented as factors.

Then l1==1 gives:

        X1    X2
[1,]  TRUE FALSE
[2,] FALSE FALSE