interep (version 0.1.0)

dmcp: This function obtains the first derivative function of MCP (Minimax Concave Penalty)

Description

This function obtains the first derivative function of MCP (Minimax Concave Penalty)

Usage

dmcp(theta, lambda, gamma = 3)

Arguments

theta

a coefficient vector.

lambda

the tuning parameter.

gamma

the regularization parameter in MCP (Minimax Concave Penalty). It balances between the unbiasedness and concavity of MCP.

Value

the first derivative of MCP function.

Details

Rigorously speaking, the regularization parametre \(\gamma\) needs to be obtained via a data-driven approach. Published studies suggest experimenting with a few values, such as 1.8, 3, 4.5, 6, and 10, then fixing its value. In our numerical study, we have examined this sequence and found that the results are not sensitive to the choice of value of \(\gamma\), and set the value at 3. In practice, to be prudent, values other than 3 should also be investigated.

Examples

# NOT RUN {
theta=runif(20,-5,5)
lambda=1
dmcp(theta,lambda,gamma=3)
# }