What is it that you want to do?  Can you formulate your question more precisely?  Also, What is your concern about correspondence analysis?  If you can do these, then we may be able to help you.

@ Ravi Vardhan Sir,  I have pool of major oxides and trace element data(of my own and also some from published data) from the dyke swarms of my study area. I have to discriminate them on the basis of their composition.I want to use  correspondence analysis  for this.

Amiya,

Still your question is not clear for me. Are you interested to discriminate tectonic setting on the basis of chemical composition? If is it like that you might be test our Multidimensional discrimination diagrams, I think, in some cases of mafic dyke, it will work. As I already tested Brazilian Mafic dykes i.e., Curaçá and Chapada. It works well in these cases.

Sanjeet Sir

• For tectonic discrimination, we have used some of your work Agrawal et al. (2004), Agrawal et al. (2008), Verma et al. (2006), Verma and Agrawal (2011) in our recent publication in IGR. Thanks to Prof. S. P. Verma sir for his support and help. However, the result is not satisfying as it indicates  IAB and MORB setting for different samples of the study area. Fact is that, the studied dykes of Eastern Dharwar Craton  are emplaced in a intraplate setting (although  same tectonic setting for ~2.37 to ~1.89 Ga time span is still a point for further work) .I think for the Paleoproterozoic Dykes, it needs a refinement. As you were telling (during our conversation at Bundelkhand), you were in a process to develop a new one for the Proterozoic dykes.Is there any work on this? Kindly let me know.
• Coming to my question, I was going through some of the publications of Klausen et al.(2010), Nilsson et al.(2013) on Petrogenesis mafic dykes, they have used Multi-variable statistical discrimination for discriminating among compositionally different swarms. I wanted to know the procedure/software used  for this purpose.

The software you may find useful will be one that has factor or principal component analysis, cluster analysis and other classification tools. I used the following: Statistica by Statsoft (http://www.statsoft.com/Products/STATISTICA/Data-Miner), ioGas by Reflex (http://reflexnow.com/iogas/). Some geochemists use JMP (http://www.jmp.com/en_us/software/jmp.html) and R - free open source software environment that works in Windows, Mac or Unix, but it will require some programming skills to build your own application. Statistica is very expensive, only Windows based and needs transferring data to Excel of other software to do custom graphics and classification. ioGas is more reasonable in price and can work both on Windows and Mac. It is more versatile for your application. It provides built-in classification diagrams for rock types by major components, SPIDER diagrams and other useful petrochemical stuff. 

llya Anisimov sir

Thank you for your reply and mentioning a number of tools for the multivariate statistical correspondence analysis.

you are welcome, Amiya. I wish you luck.

Make sure you validate and prepare data before doing any analysis. Check the data quality, lack of systematic errors, analytical methods and data origin. One need to decide weather to keep the outliers, approach values lower than detection limit, check if there is a closure of data, apply method to avoid those effects.