1. It seems the example is just for a demo purpose, but can be easily
extended to production style code.
Incorporating cross-validation does not take much efforts. For Ex.
Consider Handwritten digits data, keep all digits data except one in
training sample,
perform cross validation for various valu
w.r.t. the one-class SVM example
http://scikit-learn.org/stable/auto_examples/svm/plot_oneclass.html#example-svm-plot-oneclass-py
is there a reason why cross-validation is not used?
Also, is novelty detection with multivariate Gaussian distribution available in
sklearn?
Thank you,
--
