coefl1mstate: Obtain the coefficients

In L1mstate: L1-Regularized Multi-State Models

Description

Obtain the optimal coefficients using the first cross-validation method or the penalized cross-validation method by setting s = "lambda.min" or "lambda.pcvl". Can also obtain the coefficient at any lambda value within the range of sequence.

Usage

coefl1mstate(object,s=c("lambda.pcvl","lambda.min"))

Arguments

object	fitted cv.l1mstateR object
s	lambda value (numeric type) or cross-validation methods (character type)

Details

Return the coefficient values

Value

Return the coefficient values.

Author(s)

Xuan Dang [e-mail]

Examples

set.seed(1001)
p <- 9
times <- 1:p
rho <- 0.5
H <- abs(outer(times, times, "-"))
C <- 1 * rho^H
C[cbind(1:p, 1:p)] <- C[cbind(1:p, 1:p)] 
sigma <- matrix(C,p,p)
mu <- rep(0,p)

beta12 <- c(-.65,-.65,-.65,0,0,-.65,-.65,0,0)
beta13 <- c(-.65,-.65,0,0,0,0,-.65,0,0) 
beta23 <- c(0,-.65,-.65,0,0,-.65,-.65,0,-.65)

N <- 200
x <- mvrnorm(n=N, mu, sigma)
col_names  <-  c(sprintf("X%d", seq(1,dim(x)[2])))
colnames(x)  <-  col_names
N12  <-  N-2*N%/%5
N13 <- N-N12
hx12 <- exp(x[1:N12,] %*% beta12) 
ty12 <- rexp(N12,hx12) 
hx13 <- exp(x[(N12+1):N,] %*% beta13) 
ty13 <- rexp(N13,hx13)
hx23 <- exp(x[1:N12,] %*% beta23) 
ty23 <- rexp(N12,hx23) 
t2 <- rep(0,N)
t2[1:N12]  <-  ty12
t2[(N12+1):N]  <-  ty13
s2  <-  rep(0,N)
s2[1:N12] <-  rep(1,N12)
t3 <- rep(0,N)
t3[1:N12]  <-  ty12+ty23
t3[(N12+1):N]  <-  ty13
#censor case for state 3
tcens3 <- rbinom(n=N, prob = 0.3, size = 1)
s3 <- 1-tcens3
    
# pre-process data into long-format using msprep
dt <- data.frame(illt=t2, ills=s2, dt=t3, ds=s3, x)
tmat <- matrix(c(NA,NA,NA,1,NA,NA,2,3,NA), nrow = 3)
longdt <- msprep(time=c(NA,"illt","dt"),status=c(NA,"ills","ds"),
                 keep = col_names, data=dt,trans=tmat)

cv.l1fit <- cv.l1mstateR(longdt, nlambda = 100, nfolds = 10)
coefl1mstate(cv.l1fit, s="lambda.min")

L1mstate documentation built on Oct. 15, 2021, 5:07 p.m.