Is there an R function that does canonical correspondence analysis. Can it be done using the VR function corresp()? If not, how hard it be to write R code to do it? I am a population biologist with long but patchy programming experience in C, Smalltalk, Java and other languages. Thanks, Patrick Foley patfoley at csus.edu -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
On Fri, 16 Feb 2001, Patrick Foley wrote:> Is there an R function that does canonical correspondence analysis. Can > it be done using the VR function corresp()?What is `canonical correspondence analysis'? Correspondence analysis is a variation on canonical correlation analysis. None of my sources have `canonical correspondence analysis', but my guess is that it is a re-naming of something else.> If not, how hard it be to write R code to do it? I am a population > biologist with long but patchy programming experience in C, Smalltalk, > Java and other languages.Everything I have met so far in correspondence analysis is two or three lines of R. -- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272860 (secr) Oxford OX1 3TG, UK Fax: +44 1865 272595 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
In-Reply-To: <Pine.GSO.4.31.0102170733280.13118-100000 at auk.stats> Brian, As an ecologist, this is something I am also interested in. However, as I am more an ecologist than a statistician, I quote from the MVSP manual: "Canonical Correspondence Analysis(CCA; ter Braak, 1986,1987) is a multivariate direct gradient analysis method that has become widely used in ecology. As the name suggests, this method is derived from correspondence analysis, but has been modified to allow environmental data to be incorporated into the analysis. It is calculated using reciprocal averaging form of correspondence analysis. However, at each cycle of the averaging process, a multiple regression is performed of the sample scores on the environmental variables. New site scores are calculated based on this regression, and then the process is repeated , continuing until the scores stabilise. The result is that the axes of the final ordination, rather than simply reflecting dimensions of the greatest variability in the species data, are restricted to the linear combinations of the environmental variables and the species data. In this way these two sets of data are then directly related." Ref 1: ter Braak, CJF (1986) Canonical correspondence analysis: A new eigenfactor technique for multivariate direct gradient analysis. Ecology, 67:1167-1179 Ref 2: ter Braak, CJF (1987) The analysis of vegetation-environment relationships by canonical correspondence analysis. Vegetatio, 64:69-77 More general references are: Jongman, RHG., Ter Braak, CJF., and, van Tongeren, OFR. (1995) Data analysis in community and landscape ecology. CUP. Kent, M. & Coker, P. (1992) Vegetation description and analysis: A practical approach. Wiley. As far as I am aware, only MVSP and Canoco carry out CCA. I hope this is of some interest and it does raise the general point that programming literate ecologists don't seem to have yet fully embraced S-Plus or R. Having only recently found S-Plus and R myself I had rather hoped that some of the common ecological statistical tools such as CCA, Twinspan and Decorana, plus the miscellaneous tools described in Krebs (1989) might have found there way into an R package. Krebs, CJ (1989) Ecological Methodology, Harper & Row. Now as a 2nd edition 1995?) Cheers, Graham S -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
[Manually approved, since my filter had this looking like spam... MM] Here is some insight into the commercial value of ordination programs including canonical correspondence analysis. And it shows why a graduate student (working on bird distributions along a riparian gradient) at CSUS came to me hoping to find something less costly. Patrick Foley http://www.ptinet.net/~mjm/canoco.htm -------------- next part -------------- An HTML attachment was scrubbed... URL: https://stat.ethz.ch/pipermail/r-help/attachments/20010217/73f6585c/canoco.htm
In-Reply-To: <Pine.A41.4.10.10102170759570.18132-100000 at gpu2.srv.ualberta.ca> Barry,> If there is widespread interest in implementing Krebs as an R package, > this shouldn't be too much work.I'm not sure about wide spread interest, but as newcomer to the R/S fold I was struck that there were several biologists/ecologists who were obviously very experienced users of R. As ecologists do tend to use some procedures that are not main stream, and not commonly found in statistical programs, I thought that some of these might have made it into a R package. I would thought that the availability of a "Krebs package" or an "Ecology package" would have actually encouraged people to use R. I do know of someone who is currently looking for the equivalent of the negbinom but not limited to 180 samples. Would that restriction remain if ported to R. My remarks were more of a general nature as R seems to lend itself, to those with the skills, to pull together the miscellany of ecological data analysis tools into a single program. So although, there may not be a widespread demand for a Krebs package, I'm sure you would gather a fair number of grateful users. Cheers, Graham S -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
In-Reply-To: <5.0.0.25.1.20010218094827.03c8d080 at pop.jcu.edu.au> Richard,>Brian McArdle and students at Auckland University are working on this >(i.e. a general ecological stats module for R).That is good news, I look forward to it. I will be particularly interested to see the "hallowed" techniques that don't make it. Thanks, for passing this on, it will spur me on to keep learning "R". Cheers, Graham S -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Canonical Correspondence Analysis is an ordination method widely used
among ecologists (ter Braak). This method allow to analyze two tables
(L and E) which have the same number of rows. It is a correspondence
analysis of table L with linear constraint from E (multiple linear
regression) : the row score are linear combination of variables of E.
The columns scores are obtained by weighting averaging. CCA finds
linear combination of external variables (E) to obtain a row score
which maximizes the between-variance of columns score (obtained by
averaging) . Ter Braak called it "canonical" correspondence analysis
because it analyses two tables (like canonical analysis) but we
preferred the more explicit name of correspondence analysis with
respect to instrumental variables".
But .... in a theoretical point of view, CCA could be considered as a
canonical correlation analysis (the same links as between
correspondence analysis and canonical correlation analysis). I m just
writing a R function to perform Correspondence analysis with or
without linear contstraints (CA, CCA on row or column, Double CCA on
rows and columns) and to make the graphical representation (bi or
triplot).
I will be happy to communicate this function to anyone which is intersted in.
Sincerly,
--
St?phane DRAY
---------------------------------------------------------------
Biom?trie et Biologie ?volutive - Equipe "?cologie Statistique"
Universite Lyon 1 - Bat 711 - 69622 Villeurbanne CEDEX - France
Tel : 04 72 43 27 56 Fax : 04 78 89 27 19
04 72 43 27 57 E-mail : dray at biomserv.univ-lyon1.fr
---------------------------------------------------------------
ADE-4 http://pbil.univ-lyon1.fr/ADE-4/ADE-4F.html
---------------------------------------------------------------
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
> Date: Mon, 19 Feb 2001 10:28:40 +0100 > To: r-help at stat.math.ethz.ch > From: Stephane Dray <dray at biomserv.univ-lyon1.fr> > Subject: Re: [R] canonical correspondence analysis > > Canonical Correspondence Analysis is an ordination method widely used > among ecologists (ter Braak). This method allow to analyze two tables > (L and E) which have the same number of rows. It is a correspondence > analysis of table L with linear constraint from E (multiple linear > regression) : the row score are linear combination of variables of E. > The columns scores are obtained by weighting averaging. CCA finds > linear combination of external variables (E) to obtain a row score > which maximizes the between-variance of columns score (obtained by > averaging) . Ter Braak called it "canonical" correspondence analysis > because it analyses two tables (like canonical analysis) but we > preferred the more explicit name of correspondence analysis with > respect to instrumental variables". > But .... in a theoretical point of view, CCA could be considered as a > canonical correlation analysis (the same links as between > correspondence analysis and canonical correlation analysis). I m just > writing a R function to perform Correspondence analysis with or > without linear contstraints (CA, CCA on row or column, Double CCA on > rows and columns) and to make the graphical representation (bi or > triplot). > I will be happy to communicate this function to anyone which is intersted in.Can we encourage you to submit it as a package to CRAN? Sounds as if a number of people would be interested. Just one point: please avoid CCA, as it is unclear if it is canonical correspondence analysis canonical correlation analysis -- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272860 (secr) Oxford OX1 3TG, UK Fax: +44 1865 272595 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
In-Reply-To: <200102192057.MAA28224 at axe.humboldt.edu> Michael,> I haven't contributed it yet because it doesn't conform to R package > standards.I suspect, that this might be the case with several people. That is those who can cobble together something good enough for their own use, but not up to the standard for general release. It will be interesting to see what Brian McArdle et al come up with. I am a long way off being able to develop any modules, but learning quickly. Cheers, Graham S -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
In-Reply-To: <3A8E9D23.9A58A8E3 at csus.edu> Patrick, For something a little less costly, there is MVSP at www.kovcomp.co.uk Cheers, Graham S -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
> Date: Mon, 26 Feb 2001 16:37:15 +0200 > From: Jari Oksanen <jarioksa at sun3.oulu.fi> > >[...]> > * CCA which is the canonical ordination method in community ecology nowadays: > filling this gap would be really good (this is not my personal opinion, which > is more sceptical, but the general opinion). I'd glad to see the functions, if > they are available (and I think the eigenvector, or svd, version should be > favoured instead of the original power algorithm look-alike).A package by Stephane Dray that covers CCA is in preparation and will be on CRAN shortly. [...]> * Venables' & Ripley's isoMDS (and perhaps even more interestingly: sammon) > are already available, but they'd need some extra tools for community > ecologists. Most importantly, an additional package to `dist' with > ecologically more relevant distance measures: those that have a good > rank-order correlation with ecological distances with unimodal species > responses along ecological axes (`gradients'). By the way, the default > practice of using metric scaling as the starting configuration in isoMDS is > almost guaranteed to fall into local optima: I have tested with some data sets > using random starting configurations and usually managed to find much lower > stresses than the default (this is in common with most implementations of MDS > in statistical packages).Sure, but that is a feature not a bug. You search for a local minimum near the starting point. If you don't like cmdscale as a starting point you can easily supply another. There is no way to do other than local minimization, and not much reason to suppose that the answer is a particularly compelling one. -- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272860 (secr) Oxford OX1 3TG, UK Fax: +44 1865 272595 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
The package CoCoAn to perform constrained correspondencce analysis is
available at http://www.ci.tuwien.ac.at/R/ : Unix & Linux (Sources of
contributed packages )
Any comments would be useful.
--
St?phane DRAY
---------------------------------------------------------------
Biom?trie et Biologie ?volutive - Equipe "?cologie Statistique"
Universite Lyon 1 - Bat 711 - 69622 Villeurbanne CEDEX - France
Tel : 04 72 43 27 56 Fax : 04 78 89 27 19
04 72 43 27 57 E-mail : dray at biomserv.univ-lyon1.fr
---------------------------------------------------------------
ADE-4 http://pbil.univ-lyon1.fr/ADE-4/ADE-4F.html
---------------------------------------------------------------
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
>>>>> On Tue, 6 Mar 2001 10:20:37 +0100, >>>>> Stephane Dray (SD) wrote:> The package CoCoAn to perform constrained correspondencce analysis is > available at http://www.ci.tuwien.ac.at/R/ : Unix & Linux (Sources of > contributed packages ) Just as a side note because this seems to have been confused rather often lately: SOURCES for packages are for ALL platforms, i.e., also for windows and mac. Only BINARY packages are platform-dependent. Hence, if you install all tools (perl, ...) necessary to install source packages, you can install them on windows machines. The directory CRAN/src/contrib is not Unix & Linux, but for all platforms. The only difference is the following: On unix machines the tools are usually available by default, while on windows they are not. Hence, on unix-like platforms it is much more convienient and therefore common to install source packages directly while on windows it is not. Best, Fritz -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._