CASVD: Correspondence Analysis with SVD

A very stable solution in orthogonal CA can be obtained using Singular Value Decomposition to find the eigenvalues. Program CASVD imitates CANOCO input and output, so that you can use most types of CANOCO input files, and, for instance, plot output with CanoDraw.

Instead of this software, I recommend you to use my R function cca in vegan which performs correspondence analysis using SVD. In addition, it can perform constrained correspondence analysis (CCA), and has several support functions. In additon R Statistical Environment has other good implementations of CA.

• Linux versions are available as rpm packages (latest version: 30/5/2002):
  • casvd-2.5.2-2.i386.rpm binary package (55K).
  • casvd-2.5.2-2.src.rpm: source rpm (31K).
  To build casvd from sources, you need meschach numerical library, but you shouldn't need meschach for running the binaries, because it was I used to have an rpm binary of meschach here, but it got stale with Linux development, and I haven't made any new version. In fact, the whole casvd is seriouslu outdated, and if you want to get a new and maintained version of CA (and CCA and pCCA), you should switch to R and install the vegan package.
• DOS/Windows: Download the executable (casvd.zip: version 2.3.ms 16/1/98, 100K), or read manual page or version history (changes from v2.1 are that the program can handle data sets with missing species or plot names, and bug causing occasional FP errors was corrected, v2.3 has some cosmetic changes). This also is seriously outdated and unmaintained: install R and get vegan instead.