Shrinkage

Regularized regression in Rust, with composable losses, penalties, and solvers, dense and sparse matrices, and Python bindings for Gaussian lasso.

 

Shrinkage is a Rust library for fitting regression models with penalties that shrink coefficients, select features, or encourage structure. It supports linear, logistic, and multivariate regression, with lasso, ridge, elastic net, SLOPE, Group LASSO, and fusion penalties.

The library combines losses, penalties, and solvers through a typed API. It supports dense and sparse matrices through LazyMatrix and fits regularization paths by reusing previous solutions. Python bindings provide dense Gaussian lasso fitting, prediction, and convergence diagnostics and are under development.