Bringing in MLBase should also reexport fit! which would solve #13 additionally it comes with elementary KCV facility (also stratified) which can be reexported. This should allow to build a simple example to KCV Ridge Reg similar to what sebastian tried.
it does not implement a train_test_split function (in fact ScikitLearn.jl doesn't either). It seems to mix validation and test --> should be easy to imitate the native ScikitLearn functionality for train/test split.
fit!which would solve #13 additionally it comes with elementary KCV facility (also stratified) which can be reexported. This should allow to build a simple example to KCV Ridge Reg similar to what sebastian tried.train_test_splitfunction (in fact ScikitLearn.jl doesn't either). It seems to mix validation and test --> should be easy to imitate the native ScikitLearn functionality for train/test split.Note MLbase also implements a
gridtunefunction which may correspond to what valentin drafted.