use of apply in a function that accepts vector, matrix, or data.frame as an argument

Question

I'm trying to write a function that accepts (amongst other things) an argument that might be either a vector , a matrix , or a data.frame .

set.seed(101)
MyT <- seq(0, 1, 0.1)
S   <- sample(seq(0, 1, 0.01), 15, replace = T)
L1  <- sample(c(0,1), 15, replace = T)
L2  <- sample(c(0,1), 15, replace = T)
M1  <- as.matrix(L1)
M2  <- as.matrix(L1, L2)
D1  <- data.frame(L1)
D2  <- data.frame(L1, L2)

I'd like to write a function that is generic enough to accept L1 (or L2 ), M1 , M2 , D1 , or D2 as an argument, which I'll call myArg . Inside the function, I want to do something like:

sapply(MyT, function(t) { apply(D2[S > t, ], 2, sum) })

The line above works well. The issue is that I need to make it generic, like:

sapply(MyT, function(t) { apply(myArg[S > t, ], 2, sum) })

However, this code would fail, for instance, if myArg is L1 or L2 . In this particular case, because I am trying to subset a vector illegally, the error message will read incorrect number of dimensions . If I fix the subset issue, then I run into the dim(X) must have a positive length issue, because I can't use apply on a vector. And so on and so forth...

So, I've landed on trying to do the following:

sapply( MyT, function(t) { 
  if (length(dim(myArg)) == 0) sum(myArg[S > t])  # for vector
  else if (dim(myArg)[2] == 1) sum(myArg[S > t, ])  # for single-column matrix or data.frame
  else sapply(myArg[S > t, ], 2, sum)  # for multi-column matrix or data.frame
})

I expected this to work, but I'm still getting the Error in apply(myArg[S > t, ], 2, sum) : dim(X) must have a positive length when I use M2 as myArg .

I'm confused because dim(M2) yields [1] 15 2 .

Curious if anyone can shed light on why I get this error when I use M2 as myArg

Answer 1

R has an object framework. It has three of them, in fact (four if you count R6, five if you count proto). So why not make use of them?

myFunc <- function(x, ...)
UseMethod("myFunc")

# don't use apply() on data frames unless you know what you're doing
myFunc.data.frame <- function(x, S, t, ...)
sapply(x[S > t, ], myFunc.default)

myFunc.matrix <- function(x, S, t, ...)
apply(x[S > t, ], 2, myFunc.default)

myFunc.default <- function(x, S=1, t=0, ...)
sum(x[S > t])