Putting functions into a data frame

Question

It seems possible to assign a vector of functions in R like this:

F <- c(function(){return(0)},function(){return(1)})

so that they can be invoked like this (for example): F[[1]]().

This gave me the impression I could do this:

DF <- data.frame(F=c(function(){return(0)}))

which results in the following error

Error in as.data.frame.default(x[[i]], optional = TRUE) : cannot coerce class ""function"" to a data.frame

Does this mean it is not possible to put functions into a data frame? Or am I doing something wrong?

Solution 1

No, you cannot directly put a function into a data-frame.

You can, however, define the functions beforehand and put their names in the data frame.

foo <- function(bar) { return( 2 + bar ) }
foo2 <- function(bar) { return( 2 * bar ) }
df <- data.frame(c('foo', 'foo2'), stringsAsFactors = FALSE)

Then use do.call() to use the functions:

do.call(df[1, 1], list(4))
# 6

do.call(df[2, 1], list(4))
# 8

EDIT

The above work around will work as long as you have a named function.

The issue seems to be that R see's the class of the object as a function, looks up the appropriate method for as.data.frame() (i.e. as.data.frame.function()) but can't find it. That causes a call to as.data.frame.default() which pretty must is a wrapper for a stop() call with the message you reported.

In short, they just seem not to have implemented it for that class.

Solution 2

While you can't put a function or other object directly into a data.frame, you can make it work if you go via a matrix.

foo <- function() {print("qux")}
m <- matrix(c("bar", foo), nrow=1, ncol=2)
df <- data.frame(m)
df$X2[[1]]()

Yields:

[1] "qux"

And the contents of df look like:

  X1                                   X2
1 bar function () , {,     print("qux"), }

Quite why this works while the direct path does not, I don't know. I suspect that doing this in any production code would be a "bad thing".