Hi, Does anyone know the common modern way(s) to attempt multivariate analysis? I want to take a look at the findings of the paper below and use 8-12 variables combined instead of 1 by itself. I assume the relationships would be non-linear but beyond that I'm lost. Should I be using orthogonal polynomials? Splines? Regression trees? SOM? SVM? Clustering? K-nearest neighbor? What keywords or specific methods should I be searching for? Any hints/clues/links would be appreciated. A Non-Random Walk Down Canary Wharf: http://mpra.ub.uni-muenchen.de/9871/1/MPRA_paper_9871.pdf (FWIW I'm using Java, know C#, and will be looking maybe at R, Matlab, and WEKA.) Thanks, Stoxtrader