对r中的数据集中的所有变量进行排名和最小-最大范数

Question

I am wanting to transform all but one of the variable in my dataset and include these new, transformed variables into a new dataset. I imagine this should not be that difficult, however I have been running into some issues. The transformation I am trying to do is a min-max norm of the rank. I have been running the following code:

#create min-max norm function
normalize <- function(x) {
    return((x-min(x))/(max(x)-min(x)))
}

cols <- c(1:ncol(dt))
dt[cols] <- normalize(rank(dt[cols]))

However, when I run this code I receive the following error:

 Error: Can't use matrix or array for column indexing

I was wondering if someone could please help me out. Thank you in advance.

Answer 1

rank and normalize both take vectors. You can't give them data frames.

sapply and lapply are the classic utilties for apply ing function to columns of a data frame. A couple ways to do it:

# one step with an anonymous function, calling both normalize and rank
dt[cols] = lapply(dt[cols], function(x) normalize(rank(x)))

However, your error seems to suggest that your cols object is a matrix or array. So make sure that cols is what you think it is. And the methods I mention above assume that dt is a data.frame (or data.table , or tibble ). If dt is a matrix , array , or something else, they will not work as expected. Look at str(dt) to check (and perhaps edit that info in to your question if you still have problems). # two steps, one at a time dt[cols] = lapply(dt[cols], rank) dt[cols] = lapply(dt[cols], normalize)