2

I have a process that I'd like to loop through one variable at a time.

Although my process is much more complicated, I've used the below to illustrate the basic problem.

Suppose I want to build a histogram and do lots of other stuff for each variable in iris. The following accomplishes that goal:

hist(iris$Sepal.Length, main = paste("Histogram of Sepal.Length"))
hist(iris$Sepal.Width, main = paste("Histogram of Sepal.Width"))
hist(iris$Petal.Length, main = paste("Histogram of Petal.Length"))
hist(iris$Petal.Width, main = paste("Histogram of Petal.Width"))

However, my data frame is much larger, and my process much more complex. I'd like to wrap this in a loop like below (this does not work, but is how I'm envisioning in my head).

name.list <- names(iris)

for (i in 1:4) {

  print(i)
  print(name.list[i])
  print(paste0('iris$', name.list[i]))

  hist(paste0('iris$', name.list[i]), main = paste("Histogram of ", name.list[i]))

  # A bunch of other stuff I need to do with this variable
  # ...
  # ...

}

What am I missing here? How can I wrap this code to loop through one at a time?

pyll
  • 1,688
  • 1
  • 26
  • 44
  • 2
    Double square brackets gets the column by value of the thing in the brackets. So `iris[[ name.list[i] ]]` . – Spacedman Nov 30 '17 at 15:12
  • Thanks. This would definitely get me where I'm going, but the answer Joris provided looks like a cleaner and better way to do this. – pyll Nov 30 '17 at 15:22
  • @pyll my answer is actually the exact same as Spacedman gave you. I just wrote it out a bit more so it's easier to read for R novices. The key message here is the use of `[[`. Spacedman was very right about that. – Joris Meys Nov 30 '17 at 15:35
  • Yes, I agree. He gave me exactly what I was looking for, but the piece you added that I think makes it cleaner is not creating the `name.list` vector at all. Either way, thank you both for helping me get this sorted. – pyll Nov 30 '17 at 15:39

1 Answers1

3

It's easier if you treat a data.frame as a list. An object of class data.frame is a list, so you can use the list selection [[ here:

for( i in names(iris)){
  tmp <- iris[[i]]
  if(is.numeric(tmp))
    hist(tmp, main = paste("Histogram of",i))
}

See also the answers to this question : Pass a data.frame column name to a function

Joris Meys
  • 106,551
  • 31
  • 221
  • 263