| file names | descriptions |
|---|---|
| cpl_data_generation.R | generate simulated complete data set |
| functions.R | a collection of functions which are used in other files |
| diagnostics folder | R scripts used to do MCMC diagnostics, including traceplots, autocorrelation plots, running mean plots and effective sample size |
| CompleteDataSim folder | Complete data simulation includes Study 1,2 and 3 |
| MCARSim folder | MCAR simulation includes Study 4, 5 and 6 |
| MARSim folder | MAR simulation includes Study 7, 8, 9 and 10 |
| Obsfit folder | simulation only using observed data includes Study 11-16 |
In each simulation study folder of CompleteDataSim, there are two R scripts, one is used to update MCMC and the other is used to run MCMC, save results and do diagnostics. See here as an example.
For other kinds of simulation studies(MCAR,MAR and Obsfit), there are three R scripts in each study folder. See here as an example. 'data_generation_R' is used to generate data for specific setting based on 'cpl_data_generation.R'. 'mcmc_update.R' is used to set up the initial values and sample posterior in mcmc, 'mcmc_run.R' is used to run mcmc and do diagonistics.
In 'cpl_data_generation.R' file, true values can be changed here.
#---------------------------------
# Global set-up.
#---------------------------------
n=500 ##number of subject
K=4 ##number of latent classes
minQ=5 ##minimum number of tracked quarters
maxQ=45 ##maximum number of tracker quarters
eta_sim=c(0, 0.5, 1.5, 1)
beta_sim=c(-0.4, 0.5)
M_sim=c(0, -0.6, 0.6, 1.2)
v_sim=c(0.5, -0.3)
sgmr2_sim=1 ##true value of variance of b_{i}
sgm2_sim=1 ##true value of variance of epsilon_{it}
E_sim=1 ##true value of variance of e_{i}
In 'mcmc_update.R' files, initial values can be changed here.
#---------------------------
# PRIORS AND INITIAL VALUES
#--------------------------
beta_pri=10^{4}
M_pri=10^{4}
sgm2_pri=0.001
sgmr2_pri=0.001
v_pri=10^{4}
E_pri=0.001
eta_pri=10^{4}
inits=list(eta=c(0, 0.2, 1.3, 0.7),
beta=c(-0.5, 0.8),
M=c(0, 0.5, 1.4, -0.4),
v=c(0.3, -1),
sgmr2=1.5^2,
sgm2=1.5^2,
E=1.5^2,
b=rnorm(n, mean=0, sd=1.5), ##initial value of random effect b_{i}
e=rnorm(n, mean=0, sd=1.5) ##initial value of random effect e_{i}
)
See here for all simulation studies.