R/QR.lasso.admm.R

In cqrReg: Quantile, Composite Quantile Regression and Regularized Versions

QR.lasso.admm = function(X,y,tau,lambda,rho,beta,maxit)
{

if(missing(maxit)){
maxit = 200
}

n=dim(X)[1]
x=cbind(rep(1,n),X)

if(missing(beta)){
	beta=solve(t(x)%*%x,t(x)%*%y)
}

if(missing(rho)){
	rho=0.4
}

if(missing(lambda)){
	lambda=1
}


betah=QRPADMMCPP(X,y,beta,maxit,tau,rho,lambda)
beta=matrix(betah[-1])
b=betah[1]
return(list(beta=beta,b=b))
}