I searched for how to find feature weights and found this stackoverflow answer. It gives the following equation to get the weights:
w = t(model$coefs) %*% model$SV
The answer mentions that the method is for SVM with linear kernel. Can I use the same equation for radial? Why or Why not?