RMWreg_MCMC: MCMC sampler to fit a RMW AFT regression model

In catavallejos/RMWreg: Accelerated failure times survival regression using Rate Mixtures of Weibull distributions

Description

MCMC sampler to fit a RMW AFT regression model

Usage

RMWreg_MCMC(N, Thin, Burn, Time, Event, DesignMat, Mixing = "None",
  BaseModel = "Weibull", PriorCV = "Pareto", PriorMeanCV = 1.5,
  Hyp1Gam = 1, Hyp2Gam = 1, ...)

Arguments

N	Total number of iterations for the MCMC sampler. Use N>=max(4,Thin), N being a multiple of Thin.
Thin	Thining period for the MCMC sampler. Use Thin>=2.
Burn	Burn-in period for the MCMC sampler. Use Burn>=1, Burn<N, Burn being a multiple of Thin.
Time	Vector of length n containing survival times
Event	Vector of length n containing event indicators (TRUE / 1 is the event is observed, FALSE / 0 is the event is Eventored)
Mixing	Mixing distribution assigned to the (frailty) random effects. Possible values are "None", "Exponential", "Gamma", "InvGamma", "InvGauss", "LogNormal"
BaseModel	If BaseModel = "Weibull", a RMW regression is used. If BaseModel = "Exponential", a RME regression is used.
PriorCV	Type of prior assigned to the coefficient of variation of the survival times. Possible values are "Pareto", "TruncExp"
PriorMeanCV	Ellicited prior mean of the coefficient of variation (PriorMeanCV > 1). If Mixing = "InvGamma", PriorMeanCV must be below sqrt{3}. If Mixing = "InvGauss", PriorMeanCV must be below sqrt{5}. Default: PriorMeanCV = 1.5
Hyp1Gam	Shape hyper-parameter for the Gamma(Hyp1Gam, Hyp2Gam) assigned to gam.
Hyp2Gam	Rate hyper-parameter for the Gamma(Hyp1Gam, Hyp2Gam) assigned to gam.
...	Optional parameters. AR Optimal acceptance rate for adaptive Metropolis Hastings updates. It must be a positive number between 0 and 1. Default (and recommended): ar = 0.44 . StopAdapt Iteration at which adaptive proposals are not longer adapted. Use stopAdapt>=1. Default: StopAdapt = Burn. StoreChains If StoreChains = TRUE, MCMC chains of each parameter are stored in separate .txt files. (RunName argument used for file names). Default: StoreChains = FALSE. StoreAdapt If StoreAdapt = TRUE, trajectory of adaptive proposal variances (log scale) for each parameter are stored in separate .txt files. (RunName argument used for file names). Default: StoreAdapt = FALSE. StoreDir Directory where MCMC chain will be stored (only required if Store = TRUE). Default: StoreDir = getwd(). RunName Run-name to be used when storing chains and/or adaptive proposal variances in .txt files. PrintProgress If PrintProgress = TRUE, intermediate output is displayed in the console.

Value

A list containing MCMC draws for all parameters.

Author(s)

Catalina A. Vallejos [e-mail]

Examples

library(KMsurv)
data(alloauto)
n=dim(alloauto)[1]; k=2
Intercept=rep(1,times=n); x1=alloauto$type-1
DesignMat=cbind(Intercept,x1); rm(Intercept)
Time=alloauto$time; Event=alloauto$delta

Chain <- RMWreg_MCMC(N = 100, Thin = 2, Burn = 50,
                     Time, Event, DesignMat,
                     Mixing = "None", BaseModel = "Weibull",
                     PriorCV = "Pareto", PriorMeanCV = 1.5,
                     Hyp1Gam = 1, Hyp2Gam = 1)

catavallejos/RMWreg documentation built on Nov. 7, 2019, 2:01 p.m.