example of effects of C

[Sendai]

Here's an example of how changing C affects the surface chosen by SVM.

I used the polynomial kernel with Q = 3.

C = 0.01


C = 1


C = 100


C = 10000

[Sendai]

C = 1000000

[ivankeller]

Nice visualization.
We see how the classifier overfir (in-sample error goes down) when C increases.
It's like the regularization parameter but inverse effect (think it as if ).
Thanks Sendai.