[R] questions regarding the svmpath package (functions svmpath and predict)

[Francois BERENGER]

Hello,

I have two questions.

The svmpath package provides a svmpath function:
---
fit <- svmpath(xtrain, ytrain, kernel.function = radial.kernel, param.kernel = 0.8)
---
1) How to get the optimal lambda value out of this result?

The svmpath package also provides a predict function:
---
ytest <- predict(fit, xtest)
---

How to get a score (or a probability of belonging to one of the two classes)
instead of a label as the prediction output?

Thanks a lot,
Francois.