Learn R Programming

picasso (version 2.0.1)

picasso-package: picasso: Sparse Learning with Convex and Non-Convex Penalties

Description

Computationally efficient tools for fitting sparse generalized linear models with convex or non-convex penalties. Supported penalties include lasso, smoothly clipped absolute deviation (SCAD), and minimax concave penalty (MCP). Supported families are Gaussian, binomial, Poisson, square-root-lasso, and multinomial. Computation combines pathwise coordinate optimization with warm starts, active-set updates, screening rules, Proximal Newton iterations, and adaptive local linear approximation for applicable nonconvex models.

Arguments

Author

Jason Ge, Xingguo Li, Haoming Jiang, Mengdi Wang, Tong Zhang, Han Liu and Tuo Zhao
Maintainer: Tuo Zhao <tourzhao@gatech.edu>

Details

Core optimization routines are implemented in C++. Gaussian models use active-set coordinate descent with an automatic residual/covariance backend; binomial and Poisson models use active-set Proximal Newton/IRLS, square-root-lasso uses a global quadratic majorizer with active-set coordinate updates, and multinomial uses class-coupled Proximal Newton/IRLS. The R interface returns coefficient paths as Matrix objects and provides prediction, assessment, confusion-matrix, and cross-validation helpers.

See Also

picasso.