Share your functions & code

bryangoodrich

bryangoodrich

Probably A Mammal
#282
#282
While I'm at it, here is my scale01 function for scaling to the unit interval for both vectors and data frames. Preparing a neural network, this is handy (don't need it for the dummy variables above, which is also why I put that function together).

Again, nothing amazing. I'm just consolidating stuff I've found on SO, which usually just wraps an already existing function (in this case, scale).

Code: 
scale01 <- function(x) UseMethod("scale01", x)

scale01.data.frame <- function(x)
{
    m <- apply(x, 2, min, na.rm = TRUE)
    M <- apply(x, 2, max, na.rm = TRUE)
    as.data.frame(scale(x, m, (M-m)))
}

scale01.numeric <- function(x)
{
    m <- min(x)
    M <- max(x)
    as.numeric(scale(x, m, (M-m)))
}
 
bryangoodrich

bryangoodrich

Probably A Mammal
#283
#283
If I weren't so lazy, I'd add the column naming part to Jake's solution and benchmark to see how much faster that approach is. I always forget about diag. I also don't think do do assignments within other expressions, which can really shorten code! A+ to Dason for that.
 
Dason

Dason

Ambassador to the humans
#284
#284
bryangoodrich said:
I also don't think do do assignments within other expressions, which can really shorten code! A+ to Dason for that.
Click to expand...
That's not necessarily a bad thing. It's not any faster than doing the assignment in a different line of code and ultimately doing it inline makes it a little less readable in my opinion. I just wanted to keep it a one-liner since that was part of the beauty (to me) of Jake's solution.
 
You must log in or register to reply here.