I want to take multivariate data and represent it in 2-dimensional Euclidean space while minimizing distortion inherent in such a lossy projection. If there’s something on the web about how to do this, it’s proving difficult to find as there are many keyphrases that may or may not be used on such a page:
- Classical Multidimensional Scaling (There are other, non classical versions, but I’m not a New Coke kind of guy).
- Plane Fitting
- Dimension Reduction
- Matrix Approximation
- Matrix Fitting
- 2D Projection
- Singular Value Decomposition
- matrix of similarity data
There are probably more. I’ve seen many pay-to-play style sites that will sell you academic articles, but it is difficult to tell if these articles will properly address my needs…
Update: At the end of my rope I thought maybe I could find a local mathematician who could at least help point me in the right direction, so I searched on “site:unc.edu multidimensional scaling” which lead me right to a professor who has some really helpful papers on his site. It’s crazy that the best resource that I’ve found on the web is from a guy right down the street.