Understanding One Class SVM decision function output

Question

I have a one class classification problem that I have used a One Class SVM on. I printed out the decision function data and here is what I get :

[[ -3.37130521e+10]
 [  5.90432823e+13]
 [  4.73564603e+13]
 ..., 
 [  4.08249382e+11]
 [  4.68541816e+12]
 [  4.03591773e+11]]

How do I make sense of this? What do these rather huge numbers mean?

Answer 1

These are the distances of your samples to the separating hyperplane learned by the model. The predict method uses these by interpreting a postitive distance as +1 and a negative distance as -1.