shape of coef_ attribute

[olafwied]

I've been using the RVC model for binary classification and the shape of the coef_ and the active_ attribute after fitting is (1,n_samples) and not (n_features) what I had expected. What does the coef_ attribute represent here?

[AmazaspShumik]

@olafwied Internally RVC uses your data matrix of size (n_samples, n_features) to construct kernel of size (n_samples, n_samples) and uses it instead of your X. So after kernel construction number of features equals n_samples. It is similar to kernel construction by SVC (but SVC uses only Mercer kernels, while RVC can use any type of kernel).

If you want linear Model use ClassificationARD in skbayes.