Title: Reshaping data in R
Author: Hadley Wickham

Restructuring data is a common task in practical data analysis, and it
is often tedious and unintuitive. For example, to
prepare data for a lattice plot we may need a cumbersome command of the form:

d.music.df <- data.frame(
	artist = factor(rep(d.music[,1],5)),
 	y = as.vector(as.matrix(d.music[,3:7])),
 	meas = factor(rep(1:5, rep(62,5)), 
	labels = names(d.music[,-c(1,2)]))
)

This command takes a data matrix with five variables (columns) and
reshapes it as single column of numbers (y) and labels (meas, artist).
The code is not easy to read, difficult to explain to students, and it
is cumbersome to create. This type of operation, where we reshape the
data matrix, is extremely important and common to many analyses. Data
often has multiple levels of grouping (nested treatments, split plot
designs, or repeated measurements) and often requires investigation at
multiple levels. For example, from a long term clinical study we may
be interested investigating relationships over time, or between times,
between patients or between treatments.  Peforming these
investigations fluently requires the data to be reshaped in different
ways.

Currently the R supplies a reshape function that can perform some of
these tasks, but confounds multiple steps in the process and is hard
to use.  We propose a new conceptual framework for reshaping
operations and an R package to "deshape" data frames and then flexibly
"reshape" them to meet your needs. This framework also produces
contingency tables, cross-tabulations, summary statistics, and
provides a natural link to graphical methods such as parallel
coordinate plots, scatterplot matrices, mosaic plots and trellis
graphics.