Ploting decision boundary from any classifier in n-dimensions

The idea of this is to try to create an algorithm that samples from a hyper-plane in n-dimensions that is generated by some classification algorithm.

I than project the original data and the samples with PCA into two dimensions and plot the decision boundary.

Unlike other algorithms I've come by - this one actually samples from the hyper-plane in n-dimensions and not only predict from an already projected down to 3-d version of the data.