determine_K: Determine number of groups using information criterion

In zhan-gao/classo: Classifier-Lasso for Panel Data Grouping

Description

Determine number of groups using information criterion

Usage

determine_K(N, TT, y, X, y_raw, X_raw, lambda_seq, K_max, rho = 2/3 * (N
  * TT)^(-0.5), FUN = PLS.cvxr, beta0 = NULL, MaxIter = 500,
  tol = 1e-04)

Arguments

N	individual dimension
TT	time dimension
y	y(TN * 1)
X	X(TN * P)
y_raw	y_raw(TN * 1) raw data without standardization
X_raw	X_raw(TN * P) raw data without standardization
lambda_seq	Candidate tuning variable
K_max	Maximum number of groups
rho	Tuning parameter in the IC
FUN	choose which function to be used in PLS estimation
beta0	N*p matrix. initial estimator
MaxIter	Maximum # of iteration
tol	convergence criterion

Value

A list contains optimal K and lambda

lambda	Optimal lambda
K	Optimal K

zhan-gao/classo documentation built on April 24, 2020, 11:58 p.m.