I’m doing some experiments on some svm kernel methods.
My methodology for comparing those is having some multi-class and binary classification problems, and also, in each group, having some examples of
p > n,
n > p and
p == n.
However, finding some examples (5 or so for each of those subgroups) is really hard, so I want to generate them with
sklearn. Said so, I don’t know how to do it in a consistent and realistic way. I can generate the datasets, but I don’t know which parameters set to which values for my purpose.
So basically my question is if there is a metodological way to perform this generation of datasets, and if so, which is.