Joel Nothman
Joel Nothman
You don't think stratified K fold for classification is in textbooks? I think the argument about the over-optimistic estimate is that sampling at random should approximate the natural variation in...
The way I'd consider this is: add param return_X_y=True. When set to False, a bunch would be returned with additional descriptive properties. Cf. non-generated datasets.
Yes, although the conversion of coef_ to feature_importances_ in general requires a parameter: which norm to take to make all values non-negative.
You're welcome if you're confident you know where to get started, and if @hermidalc did not want to do it
I don't mind. It's just a limitation of our interface
I agree with Nicolas: the test is just being lazy.
I'd usually think that if we design the objective function such that the regularisation coefficient is sample size invariant, then it should be invariant to the scale of sample weights....
I think we have a general sense here that *ordinarily* the three invariances should hold. However, I think we can find a bit more clarity about the exceptions to that...
I think since people have been thinking in terms of algorithms, it might take a bit of work to reframe your thoughts in terms of parameters. But testing for invariances...
I think it's unclear what needs to happen in this issue, and I doubt they all can be fixed straightforwardly.