powThresh: Thresholding function for penalized regression with power...

In maryclare/powopt: Solving Penalized Regression Problems with Power Penalty by Coordinate Descent

Thresholding function for penalized regression with power penalty

Description

Gives the value of β that minimizes:

(β - z)^2/2 + λ|β|^q

for fixed z, λ > 0 and q > 0.

For q ≤ 1, uses thresholding function given in Marjanovic and Solo (2014). For q > 1, computed bisection.

Usage

powThresh(z =  1, lambda = 1, q = 1)

Arguments

z	vector of values to find the minimizing beta for (powThresh is applied coordinatewise when length(z) > 1)
lambda	scalar multiplier applied to penalty
q	scalar power of penalty function

Source

For q≤ 1, uses thresholding function given in Marjanovic and Solo (2014). Marjanovic, Goran, and Victor Solo. "l_q Sparsity Penalized Linear Regression With Cyclic Descent." IEEE Transactions on Signal Processing 62.6 (2014): 1464-1475.

maryclare/powopt documentation built on Feb. 5, 2023, 6:05 p.m.