[R] Elegant way to transform dataframe?

Barry Rountree rountree at uga.edu
Sun Nov 23 19:00:15 CET 2008

---- Original message ----
>Date: Thu, 20 Nov 2008 09:41:11 +0100
>From: "David Hajage" <dhajage.r at gmail.com>  
>Subject: Re: [R] Elegant way to transform dataframe?  
>To: "Barry Rountree" <rountree at uga.edu>
>   I suggest you to have a look to the reshape package,
>   by Hadley Wickham : http://had.co.nz/reshape/

That's exactly what I needed -- thanks!  

A few comments for other people who might be interested in trying it:

1)  "reshape" is the name of the package, "melt" and "cast" are the commands.  There's a "reshape" command that's part of the stats package and it does something similar, but it's not nearly so general.

2)  I didn't find the package documentation helpful, but the website has links to several papers that provide much better examples.

3)  The package requires an up-to-date version of R.  I had been using an old version and trying to install the package gave a "not found" error.  It wasn't until I downloaded the package and tried a local install that I got a useful error message about the version mismatch.  

Thanks again,


>   2008/11/20 Barry Rountree <rountree at uga.edu>
>     Hello,
>     I have a dataframe with columns like:
>      parameter1 parameter2 metricname value
>     and I'd like to transform it into a dataframe with
>     columns:
>      parameter1 parameter2 metric1 metric2 metric3
>     where the values for each metric would be in the
>     appropriate column.  There are often multiple
>     values for a given (parameter1, parameter2,
>     metricname) triple.  In this case, I want to use
>     the minimum value.
>     A constant number of metrics is associated with
>     each possible combination of parameters.
>     I've cobbled together code that performs this
>     using nested loops and far too many variables.
>      The number of metrics has now gone up by an order
>     of magnitude and I'm looking for a more elegant
>     way of doing this.
>     If anyone has a short bit of code that can deal
>     with this, I'd really appreciate having a look.
>     Thanks much,
>     Barry
>     ______________________________________________
>     R-help at r-project.org mailing list
>     https://stat.ethz.ch/mailman/listinfo/r-help
>     PLEASE do read the posting guide
>     http://www.R-project.org/posting-guide.html
>     and provide commented, minimal, self-contained,
>     reproducible code.

More information about the R-help mailing list