XW, you need to do 0-1 encode for categorical or dummy variables.

On 7/24/07, Xin Wei <xinwei@stat.psu.edu> wrote: > I am trying to perform discriminant analysis to predict which of two > groups a individual belong to. However, my understanding is that proc > discrim only use continuous varialbe to construct discriminant function. > But my separation variables are all binary. What procedure I ought to use > to do the discriminant analysis with all binary variables? > > thanks >

-- =============================== "I am dying with the help of too many physicians." - Alexander the Great, on his deathbed =============================== WenSui Liu (http://spaces.msn.com/statcompute/blog) ===============================