Question

I have a 2D contingency table in R; it is a table object. I want to transform it into a new table by applying a function on each of its elements.
I looked at sapply, tapply, etc., but they are all aimed at summarising/aggregating the data. I’ve written my own mapping function which does this, which I reproduce below:

map.table = function(t,fn)
{
    rows = dim(t)[1]
    columns = dim(t)[2]
    x = matrix(nrow=rows, ncol=columns)

    rownames(x) = unlist(dimnames(bush.t)[1], use.names=FALSE)
    colnames(x) = unlist(dimnames(bush.t)[2], use.names=FALSE)
    for(row in seq(from=1, to=rows))
    {
        for(column in seq(from=1, to=columns))
        {
            x[row,column] = fn(t[row,column])
        }
    }
    as.table(x)
}

This creates a matrix from scratch, fills up the dimension names, and the elements. Is there a better way of doing this? Is there an R function/package which already does this functionality?

Answer 1

You can probably just use apply:

set.seed(21)
x <- data.frame(a=sample(letters[1:5],20,TRUE),
                b=sample(letters[1:5],20,TRUE))
y <- table(x)
z <- as.table(apply(y, 1:2, sqrt))