Creating new named variable in dataframe using loop and naming convention

Question

One of the most frustrating things about R is the difficulty of creating new dataframe variables using names, algorithmically and intuitively.

Suppose I have a dataframe with some variables, and want to create new variables based on them in a loop. For example, I want to create new variables which are the cumulative sum of existing variables, and named df$var_cumul

temp<-as.data.frame(cbind(seq(0:10),seq(10:20)))
names(temp)<-c("x","y")
for (i in 1:ncol(temp)) {
  vname<-names(temp)[i]
  assign(paste("temp$",vname,"_cumul",sep=""),cumsum(contrs[,i]))
}

No permuation of that I've found works. This is probably one of my biggest issue with R on a regular basis.

Is there an easy intuitive way to do this?

You need to study the help page for "[[" more carefully. The key is avoidance of the futile effort to make $<- do something it cannot do. It would also help if you unlearned the abomination of as.data.frame(cbind(...)) — IRTFM
– IRTFM, Commented Jan 14, 2013 at 18:47
.. also never ever use 1:ncol, 1:nrow or 1:length ... (it doesn't do what you think it does and such cases are a pain to debug ...) - you want to use seq_len() and friends instead. — Simon Urbanek
– Simon Urbanek, Commented Jan 14, 2013 at 18:55
With DWin's second point, I believe he's referring to the fact that data.frame(x = 1:11, y = 1:11) is (among other things) a whole lot less typing. I would seriously recommend playing around with the output of things like seq(10:20) because that's pretty horrible too. (Reading ?seq would be a good start.) — joran
– joran, Commented Jan 14, 2013 at 19:05
My point was that cbind coerces to a matrix, which removes all attributes and makes all columns the "lowest possible denominator" generally a character matrix. — IRTFM
– IRTFM, Commented Jan 14, 2013 at 19:08
DWin - I'd say that I'd need to find the "[[' page first! It didn't come up in any search or reference about achieving this, and I'm sorry to say I'm not going to read every reference manual page. Hopefully I'll be able to reference this from now on. Much thanks! — Daniel Egan
– Daniel Egan, Commented Jan 14, 2013 at 19:09

Simon Urbanek · Accepted Answer · 2013-01-14 18:50:49Z

7

Yes, the [[ operator:

temp <- data.frame(x = 0:10, y = 10:20)

for (n in names(temp))
  temp[[paste0(n, "_cumsum")]] <- cumsum(temp[[n]])

answered Jan 14, 2013 at 18:50

Simon Urbanek

14k48 silver badges45 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Arun Raja Over a year ago

If I want to create 3 dataframe can I create it in a for loop? my case is for(i in 1:3){ df.i <- data.frame(something) }

Collectives™ on Stack Overflow

Creating new named variable in dataframe using loop and naming convention

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related