[LRNRQ] Add <algorithm> from package fuser

[tdhock]

Algorithm

Regularized linear model.

Package

https://cloud.r-project.org/web/packages/fuser/

Supported types

• classif
• clust
• dens
• [ x] regr
• surv

I have checked that this is not already implemented in

• [x ] mlr3
• [x ] mlr3learners
• [x ] mlr3extralearners
• Other core packages (e.g. mlr3proba, mlr3keras)

Why do I think this is a useful learner?

There are N rows of data and G groups, each of which has a corresponding subset of rows. G linear models are fit, with the coefs brought together using fused lasso. Interpretable: linear model coefs are the same across some pairs of groups, but different across others.

Further Optional Comments

@EngineerDanny has written a mlr3 Learner for this, and the mlr3 team invited us to submit it as a learner in this package, so can you please submit a PR?