scalescad_mcp: The Scaled MCP or SCAD

In chensyustc/SC19027: The R package for StatComp Course by 19027

Description

The scaled penalized methods with the minimax concave penalty(MCP) or smoothly clipped absolute deviation penalty(SCAD).

Usage

scalescad_mcp(X, y, lam0, method)

Arguments

X	X is an n by p design matrix where n is the sample size and p is the data dimension.
y	Y is the n dimensional response vector.
lam0	the initial value of penalty level.
method	the penalized methods,can be "scad" or "mcp".

Value

hsigma the estimated noise level

coefficients the estimated coefficients

Examples

## Not run: 
n<-200;p<-2000
data<-datageneration(n,p,0)
X<-data$X
Y<-data$Y
lam0<-sqrt(2*log(p)/n)
mcp<-scalescad_mcp(X,Y,lam0,method="mcp")
sigma<-mcp$hsigma

## End(Not run)

chensyustc/SC19027 documentation built on Jan. 3, 2020, 9:17 p.m.