On 4 Mar 2004 at 15:07, Gang Chen wrote: > I'm thinking to adopt the regression approach to run analysis of > variance. It seems that there are a lot of issues involved in terms of > the ways to implement the approach computationally, such as > reparameterization through different ways of coding in dummy > variables, model reduction through constraints, etc.. Do you happen to > know any available C or Matlab routines for this? And can you > recommend some books on this kind of computation algorithms? > > Thanks, > Gang Chen > > .
A very good system to do this flexibly is R (google for CRAN) or the commercial S-Plus. This will give full liberty in choice of parametrization. For a discussion of the modelling issues underlying choice of parameterization, there is no better guide than W.N. Venables & B.D. Ripley: "Modern Applied Statistics with S". (especially chapter 6) Kjetil Halvorsen > . > ================================================================= > Instructions for joining and leaving this list, remarks about the > problem of INAPPROPRIATE MESSAGES, and archives are available at: . > http://jse.stat.ncsu.edu/ . > ================================================================= . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
