Unified Algorithm for Non-convex Penalized Estimation for Generalized Linear Models

An efficient unified nonconvex penalized estimation algorithm for Gaussian (linear), binomial Logit (logistic), Poisson, multinomial Logit, and Cox proportional hazard regression models. The unified algorithm is implemented based on the convex concave procedure and the algorithm can be applied to most of the existing nonconvex penalties. The algorithm also supports convex penalty: least absolute shrinkage and selection operator (LASSO). Supported nonconvex penalties include smoothly clipped absolute deviation (SCAD), minimax concave penalty (MCP), truncated LASSO penalty (TLP), clipped LASSO (CLASSO), sparse ridge (SRIDGE), modified bridge (MBRIDGE) and modified log (MLOG). For high-dimensional data (data set with many variables), the algorithm selects relevant variables producing a parsimonious regression model. Kim, D., Lee, S. and Kwon, S. (2018) , Lee, S., Kwon, S. and Kim, Y. (2016) , Kwon, S., Lee, S. and Kim, Y. (2015) . (This research is funded by Julian Virtue Professorship from Center for Applied Research at Pepperdine Graziadio Business School and the National Research Foundation of Korea.)


Travis-CI Build Status ncpen package fits the generalized linear models with various nonconvex penalties. Supported regression models are Gaussian (linear), binomial Logit (logistic), multinomial Logit, Poisson and Cox proportional hazard. A unified algorithm is implemented based on the convex concave procedure and the algorithm can be applied to most of the existing nonconvex penalties. The algorithm also supports convex penalty: least absolute shrinkage and selection operator (LASSO). Supported nonconvex penalties include smoothly clipped absolute deviation (SCAD), minimax concave penalty (MCP), truncated LASSO penalty (TLP), clipped LASSO (CLASSO), sparse ridge (SRIDGE), modified bridge (MBRIDGE) and modified log (MLOG). This package accepts a design matrix X and vector of responses y, and produces the regularization path over a grid of values for the tuning parameter lambda. Also provides user-friendly processes for plotting, selecting tuning parameters using cross-validation or generalized information criterion (GIC), l2-regularization, penalty weights, standardization and intercept. For a data set with many variables (high-dimensional data), the algorithm selects relevant variables producing a parsimonious regression model.

Related research paper can be found at ncpen paper. A recent manual is avaialbe at ncpen manual and for an example use, see ncepn example.

(This research is funded by Julian Virtue Professorship from Center for Applied Research at Pepperdine Graziadio Business School and the National Research Foundation of Korea.)

Authors

Dongshin Kim, Sunghoon Kwon, Sangin Lee

References

News


title: "NEWS" author: "Dongshin Kim" date: "September 3, 2018" output: html_document

Update news (11/16/2018)

  1. Multinomial Logit and Cox proportional hazard models are added.
  2. Significant performance improvements.
  3. Minor bug fixes.
  4. Paper is available at http://arxiv.org/abs/1811.05061.

ncpen R package (2/19/2018)

We are releasing ncpen R pakcage: non-convex penalty estimation. Any comnents are welcome.

URL: https://github.com/zeemkr/ncpen Bug Reports: https://github.com/zeemkr/ncpen/issues

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("ncpen")

1.0.0 by Dongshin Kim, 8 months ago


https://github.com/zeemkr/ncpen


Report a bug at https://github.com/zeemkr/ncpen/issues


Browse source code at https://github.com/cran/ncpen


Authors: Dongshin Kim [aut, cre, cph] , Sunghoon Kwon [aut, cph] , Sangin Lee [aut, cph]


Documentation:   PDF Manual  


GPL (>= 3) license


Imports Rcpp

Linking to Rcpp, RcppArmadillo


See at CRAN