For example, let's say I want 100 variables to predict the likelihood of that sample being in 4 different categories A, B, C, or D. Do I set up 4 columns each corresponding to one of those categories with a value of 1 as the correct category and a value of 0 being not the correct one, and then run 4 different searches like so:
A = logistic(f0()+f1()+f2()*f(v1,v2,v3...) B = logistic(f0()+f1()+f2()*f(v1,v2,v3...)
C = logistic(f0()+f1()+f2()*f(v1,v2,v3...)
D = logistic(f0()+f1()+f2()*f(v1,v2,v3...)
and then see whichever has the highest probability?
I don't know if this is correct because most of the outputs when I do it this way are 0 or 1 rather than somewhere between.