Lorenz Schmidt

sparse PCA depends on #46

Perhaps, you can take a look at the mnist crate and implement something similar for the face dataset? There is also an open issue here to replace the downloader: https://github.com/davidMcneil/mnist/pull/8

continued in #242

all of your items are hyperparameters, they are fixed prior to the training. The model parameter would be the eigen-decomposition. https://machinelearningmastery.com/difference-between-a-parameter-and-a-hyperparameter/ It would be awesome if you can re-use parts...

awesome :tada: the current implementation of elasticnet only supports a single target, so you have to solve for every `beta` separately with `Y=X alpha`. Extension to multinomial regression can be...

Depends on https://github.com/rust-ndarray/ndarray-linalg/pull/275, this is already possible in the native Rust implementation but for fast LAPACK not yet

both points are right, we would just follow the ndarray model more closely. Because of the coming holidays, my plan was to schedule that for beginning of next year. You...

no this should default to `Array1`

We can't ensure that somebody is holding a reference while modifying the content. One solution would involve [RefCell](https://doc.rust-lang.org/std/cell/struct.RefCell.html), which basically moves the borrow check to the runtime and panics when...

It should return `impl Iterator`, but I don't know wether the model type can be chosen on the return type of a closure