Port to MLJ.jl

[azev77]

Hi and thank you for this package!
have you considered porting it to MLJ.jl?

[filipebraida]

Yes, we are.

However, we are currently working on a big project and I don't know if we would be able to change PLS in the short term.

@lalvim

[azev77]

Sounds good.
Btw, porting a package to MLJ usually involves adding one script to the repo.
For example: ParallelKmeans.jl does this in mlj_interface.jl.

[filipebraida]

Thanks. I'll check with @lalvim to do this. It seems to be easy to porting.

[lalvim]

Hi, I just portted the package to mlj in a branch named mlj. As soon as possible, I will merge to the master and make official. Out of curiosity, how did you discover the package? Are you part of the mlj team? Em sex., 1 de mai. de 2020 às 17:58, Filipe Braida <[email protected]> escreveu:

…

Thanks. I'll check with @lalvim <https://github.com/lalvim> to do this. It seems to be easy to porting. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/lalvim/PLSRegressor.jl/issues/8#issuecomment-622565731>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA65XTLBXCD7IIZZBRGMWALRPMZYTANCNFSM4MXKCWUQ> .

-- *------------------------------------------------------------------------------* *"Dave Bowman: Open the pod bay doors, HAL. * *HAL: I’m sorry Dave, I’m afraid I can’t do that." * Best Regards, D.Sc. Leandro G. M. Alvim. Professor Adjunto do Departamento de Ciência da Computação (DCC) Instituto Multidisciplinar (IM) - Campus Nova Iguaçu Universidade Federal Rural do Rio de Janeiro (UFRRJ) Skype: lgmalvim

[azev77]

@lalvim soinds great.
I discovered MLJ, by searching google “ML in Julia...”
I discovered your package, searching PLS julia
I try to contribute to MLJ a little when I can.

Check out my script to automatically run all appropriate MLJ models:

https://discourse.julialang.org/t/custom-xgboost-loss-function-w-zygote-julia-computing-blog-post/35811/13?u=albert_zevelev

[azev77]

@ablaom @tlienart check out this package. They are working on an MLJ interface.

[lalvim]

Hi, We are finalizing the validation of the mlj branch. We decided to set the interface as the default. The PLS algorithm is one of the best regression algorithms for few data and its implementation is unusual. For example, we can not find this in scikit. The same is true for the non-linear version. We hope that one day it can be integrated into MLJ. We also have another package called perceptrons (https://github.com/lalvim/Perceptrons.jl) that contains several versions (voted, averaged, ...) of the perceptron algorithm. This is yet to be adapted to the MLJ interface. It is also not common in libraries. Em qua., 3 de jun. de 2020 às 00:26, azev77 <[email protected]> escreveu:

…

@ablaom <https://github.com/ablaom> @tlienart <https://github.com/tlienart> check out this package. They are working on an MLJ interface <https://github.com/lalvim/PLSRegressor.jl/blob/mlj/src/mlj_interface.jl>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/lalvim/PLSRegressor.jl/issues/8#issuecomment-637930173>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA65XTLRFIXR3QGWBYQSBY3RUW7GLANCNFSM4MXKCWUQ> .

-- *------------------------------------------------------------------------------* *"Dave Bowman: Open the pod bay doors, HAL. * *HAL: I’m sorry Dave, I’m afraid I can’t do that." * Best Regards, D.Sc. Leandro G. M. Alvim. Professor Adjunto do Departamento de Ciência da Computação (DCC) Instituto Multidisciplinar (IM) - Campus Nova Iguaçu Universidade Federal Rural do Rio de Janeiro (UFRRJ) Skype: lgmalvim

[tlienart]

Hello! great that you'd like to use MLJ as the default interface and awesome for PLS; I'd be happy if we could run some comparisons with the more classic models in MLJLinearModels.jl at some point.

I have a fair few comments for the MLJ interface (e.g.: you don't need to using MLJ), would you guys mind

1. adding me as a collaborator
2. opening a PR from your #mlj branch so that I can comment on it?

thanks!

[filipebraida]

Thank you for your help!
I created the PR and added you as a reviewer.

[ablaom]

@filipebraida I think this can be closed, no?