Add penalty factors & multinomial regression support

[cyradil]

Please add support for multinomial regression in the newest version of pyglmnet.

Please also add support for penalty factors (regularization penalties per feature) as described here: https://glmnet.stanford.edu/articles/glmnet.html#other-function-arguments-1.

These are supported in the Python wrapper versions of glmnet and would provide great benefit here as well.

[jasmainak]

hi @cyradil thanks for your interest. Are you using pyglmnet in your work currently?

Adding these features might require some work. Do you want to help? We did have multinomial regression in the past, so it might be possible to revamp them.

[cyradil]

Hi @jasmainak. I am hoping to use it for my work. I am happy to share if I make any progress. My 3 attempts so far have failed sadly. I am newer to object-oriented programming and struggle to properly track variable inheritances haha!