MLJLinearModels.jl
MLJLinearModels.jl copied to clipboard
JuliaAI
Generalized Linear Regressions Models (penalized regressions, robust regressions, ...)
``` julia> enet = ElasticNetRegression() GeneralizedLinearRegression{L2Loss, CompositePenalty} loss: L2Loss L2Loss() penalty: CompositePenalty fit_intercept: Bool true penalize_intercept: Bool false scale_penalty_with_samples: Bool true julia> ElasticNetRegressor(enet) ERROR: MethodError: no method matching ElasticNetRegressor(::GeneralizedLinearRegression{L2Loss, CompositePenalty})...
This PR is meant to extend the package to return output with matching typing of the given input data. Specifically I am adding the flexibility to all functions to handle...
Is there a motivation for [hardcoding](https://github.com/JuliaAI/MLJLinearModels.jl/blob/dev/src/loss-penalty/generic.jl#L21) the scaling in the loss penalties as `Float64`? If not, would a more generic definition allow for multiple types of regression outputs (Float64, 32,...
a la http://www.seas.ucla.edu/~vandenbe/133A/lectures/mols.pdf probably just a matter of adding a dispatch for `y::AbstractMatrix` and stacking the columns but first wanted to see if there is receptiveness to making multiobjective OLS...
what would it take to support box / positive constraints on the Lasso / ElasticNet solvers? is this compatible with the existing API, and if so where could I get...
there are a few other regularized forms of Lasso / ElasticNet I would find useful, like group lasso, trend filtering, generalized lasso (& generalized Tikhonov). is this compatible with existing...
I think we could reduce learning curves by including some cut and past examples for the various regressors... it would also be good to include some discussion of when one...
To future proof, I suggest adding these. Here's an example: https://github.com/JuliaAI/MLJXGBoostInterface.jl/blob/4456438d134c20207c0c8383afb14d9f7bfa816c/test/runtests.jl#L205
Following discussion in https://github.com/JuliaAI/MLJLinearModels.jl/pull/138 (https://github.com/JuliaAI/MLJLinearModels.jl/pull/138#issuecomment-1411411410) there shouldn't be an error if ISTA/FISTA is used with `gamma=0` but it'd be good to check. Might be even better to just have a...
This issue is closely related to https://github.com/alan-turing-institute/MLJLinearModels.jl/issues/14 I'm trying to compare performance of `fit!` and `evaluate!` in Julia and analogous methods in scikit-learn in Python. The code and data are...
