R help - Nov 2010 - exploratory analysis of large categorical datasets

Dear List,


I am looking to perform exploratory analyses of two (relatively) large
datasets of categorical data. The first one is a binary 80x100 matrix, in
the form:


matrix(sample(c(0,1),25,replace=TRUE), nrow = 5, ncol=5, dimnames = list(c(
"group1", "group2","group3",
"group4","group5"), c("V.1", "V.2",
"V.3",
"V.4", "V.5")))


and the second one is a multistate 750x1500 matrix, with up to 15
*unordered* states per variable, in the form:


matrix(sample(c(1:15),25,replace=TRUE), nrow = 5, ncol=5, dimnames = list(c(
"group1", "group2","group3",
"group4","group5"), c("V.1", "V.2",
"V.3",
"V.4", "V.5")))


Specifically, I am looking to see which pairs of variables are correlated.
For continuos data, I would use cor() and cov() to generate the correlation
matrix and the variance-covariance matrix, which I would then visualize with
symnum() or image(). However, it is not clear to me whether this approach is
suitable for categorical data of this kind.


Since I am new to R, I would greatly appreciate any input on how to approach
this task and on efficient visualization of the results.


Many thanks in advance,

Lara

	[[alternative HTML version deleted]]

Hi:

A good place to start would be package vcd and its suite of demos and
vignettes, as well as the vcdExtra package, which adds a few more goodies
and a very nice introductory vignette by Michael Friendly. You can't fault
the package for a lack of documentation :)

You might also find the following link useful:  http://www.datavis.ca/R/
Scroll down to 'vcd and vcdExtra', and further down to
'tableplot', which
was recently released on CRAN.

HTH,
Dennis

On Thu, Nov 11, 2010 at 2:09 PM, Lara Poplarski
<larapoplarski@gmail.com>wrote:
> Dear List,
>
>
> I am looking to perform exploratory analyses of two (relatively) large
> datasets of categorical data. The first one is a binary 80x100 matrix, in
> the form:
>
>
> matrix(sample(c(0,1),25,replace=TRUE), nrow = 5, ncol=5, dimnames = list(c(
> "group1", "group2","group3",
"group4","group5"), c("V.1", "V.2",
"V.3",
> "V.4", "V.5")))
>
>
> and the second one is a multistate 750x1500 matrix, with up to 15
> *unordered* states per variable, in the form:
>
>
> matrix(sample(c(1:15),25,replace=TRUE), nrow = 5, ncol=5, dimnames >
list(c(
> "group1", "group2","group3",
"group4","group5"), c("V.1", "V.2",
"V.3",
> "V.4", "V.5")))
>
>
> Specifically, I am looking to see which pairs of variables are correlated.
> For continuos data, I would use cor() and cov() to generate the correlation
> matrix and the variance-covariance matrix, which I would then visualize
> with
> symnum() or image(). However, it is not clear to me whether this approach
> is
> suitable for categorical data of this kind.
>
>
> Since I am new to R, I would greatly appreciate any input on how to
> approach
> this task and on efficient visualization of the results.
>
>
> Many thanks in advance,
>
> Lara
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help@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.
>
	[[alternative HTML version deleted]]