A unified algorithm for the non-convex penalized estimation: The ncpen package

Abstract

Various R packages have been developed for the non-convex penalized estimation but they can only be applied to the smoothly clipped absolute deviation (SCAD) or minimax concave penalty (MCP). We develop an R package, entitled ncpen, for the non-convex penalized estimation in order to make data analysts to experience other non-convex penalties. The package ncpen implements a unified algorithm based on the convex concave procedure and modified local quadratic approximation algorithm, which can be applied to a broader range of non-convex penalties, including the SCAD and MCP as special examples. Many user-friendly functionalities such as generalized information criteria, cross-validation and ℓ 2 -stabilization are provided also.

Folded concave penalized learning of high-dimensional MRI data in Parkinson’s disease

“…We propose a grid search for tuning parameters λ and ρ while fixing the value of a to reduce computation complexity. Here we fixed a = 0.01 and obtained relatively good numerical results, and we refer readers to Zhang (2010) and Kim et al (2018) for guidance and heuristics for the selection of tuning parameters. Then we calculated AUC and accuracy scores using the test dataset that included the remaining 70 subjects.…”

Targeted Random Projection for Prediction From High-Dimensional Features

“…This proves the second part of ( 9), and the following the same procedure as in Theorem 3 the proof is completed. 9 An Addition to Section 4.1 of the Paper 9.1 Details of specifications of the tuning parameters of the competitors SCAD and MCP are calculated using three packages, viz., SIS, ncvreg and ncpen (Kim et al, 2018). For one-step SCAD (1-SCAD) we use a R-code provided by the authors of Fan et al (2014).…”