CODEWITHSUNDEEP
rdataframe
Gregor Thomas
Gregor Thomas

Reputation: 145745

How can I repeat a data.frame?

It's easy to repeat a data.frame once,

mt2 <- rbind(mtcars, mtcars)

But what's an R-like way to do this generally? If I want 10 copies of mtcars together I could

mt10 <- mtcars
for (i in 2:10) mt10 <- rbind(mt10, mtcars)

which is plenty concise, but seems not in the spirit of R. Is there a better way to do this, or a clever trick using vector recycling?

Upvotes: 4

Views: 2057

Answers (4)

David Rubinger
David Rubinger

Reputation: 3938

dplyr-based solution (from here)

library(dplyr)
mtcars %>% slice(rep(1:n(), 2))  # 2 copies of data frame

Upvotes: 1

dardisco
dardisco

Reputation: 5274

If you can tolerate another package: 

require(mefa)
rep(mtcars,10)
  • works after all!

It appears a little faster:

system.time(mtcars[rep(1:nrow(mtcars),1e5),])
system.time(mtcars[rep(seq_len(nrow(mtcars)),1e5),])
system.time(rep(mtcars,1e5))

Gives:

 user  system elapsed 
 17.41    0.19   17.89
 17.11    0.31   17.91
 6.79    0.70    7.67

Caveats: 1) rep will not reproduce the row names; this is a separate step. 2) mefa::rep(mtcars,10) doesn't work as: Error: 'rep' is not an exported object from 'namespace:mefa'. Also mefa:::rep(mtcars,10) gives object 'rep' not found. Not sure why this is...

Upvotes: 3

Matthew Plourde
Matthew Plourde

Reputation: 44614

@joran's answer is very elegant and you should use it if duplicating the rownames isn't required. However, this way will also duplicate the rownames:

do.call(rbind, replicate(10, mtcars[1:10, ], simplify=FALSE))

Upvotes: 6

joran
joran

Reputation: 173517

Here's a very simple method:

mtcars[rep(1:nrow(mtcars),2),]

or using better "grammar":

mtcars[rep(seq_len(nrow(mtcars)),2),]

As GSee notes below, one difference here is that rbind will replicate the row names exactly, while using indexing will force unique row names by appending digits. Off the top of my head, I think the only fix would be to set the row names (again using rep) after the fact.

Upvotes: 8

Related Questions