bayesSurv (version 2.6)

files2coda: Read the sampled values from the Bayesian survival regression model to a coda mcmc object.

Description

This function creates a coda mcmc object from values found in files where sampled values from bayessurvreg1 function are stored or from data.frames.

Usage

files2coda(files, data.frames, variant = 1, dir = getwd(), start = 1, end, thin = 1, header = TRUE, chain)

Arguments

files

a vector of strings giving the names of files that are to be converted to coda objects. If missing and data.frames is also missing, all appropriate files found in a directory dir are converted to coda objects. File "iteration.sim" is always used (if found) to index the sampled values. If this file is not found the sampled values are indexed from 1 to the sample size. If "mixture.sim" appeares here, only the column with number of mixture components is converted to the coda object.

data.frames

a vector of strings giving the names of data.frames that are to be converted to coda objects.

variant

a variant of bayessurvreg function used to generate sampled values. This argument is only used to identify appropriate files when files argument is missing.

Currently only 1 is supported to cooperate with bayessurvreg1.

dir

string giving the directory where it will be searched for the files with sampled values.

start

the first row (possible header does not count) from the files with the sampled values that will be converted to coda objects.

end

the last row from the files with the sampled values that will be converted to coda objects. If missing, it is the last row in files.

thin

additional thinning of sampled values (i.e. only every thin value from files and data.frames is considered).

header

TRUE or FALSE indicating whether the files with the sampled values contain also the header on the first line or not.

chain

parameter giving the number of the chain if parallel chains were created and sampled values stored in data.frames further stored in lists(). If missing, data.frames are not assumed to be stored in lists.

Value

A list with mcmc objects. One object per file or data.frame.

Examples

## *** illustration of usage of parameters 'data.frames' and 'chain' ***
## *********************************************************************
## Two parallel chains with four variables, stored in data.frames
## data.frames are further stored in lists
library(coda)

group1 <- list();  group2 <- list();  group3 <- list()
    ## first chain of first two variables:
  group1[[1]] <- data.frame(var1 = rnorm(100, 0, 1), var2 = rnorm(100, 5, 4))
    ## second chain of first two variables:   
  group1[[2]] <- data.frame(var1 = rnorm(100, 0, 1), var2 = rnorm(100, 5, 4))
    ## first chain of the third variable:
  group2[[1]] <- data.frame(var3 = rgamma(100, 1, 1))
    ## second chain of the third variable:
  group2[[2]] <- data.frame(var3 = rgamma(100, 1, 1))
    ## first chain of the fourth variable:
  group3[[1]] <- data.frame(var4 = rbinom(100, 1, 0.4))
    ## second chain of the fourth variable:
  group3[[2]] <- data.frame(var4 = rbinom(100, 1, 0.4))  

  ## Create mcmc objects for each chain separately
mc.chain1 <- files2coda(data.frames = c("group1", "group2", "group3"), chain = 1)
mc.chain2 <- files2coda(data.frames = c("group1", "group2", "group3"), chain = 2)

  ## Create mcmc.list to represent two parallel chains
mc <- mcmc.list(mc.chain1, mc.chain2)
rm(mc.chain1, mc.chain2)


## *** illustration of usage of parameter 'data.frames' without 'chain' ***
## ************************************************************************
## Only one chain for four variables was sampled and stored in three data.frames
    ## chain of first two variables:
  group1 <- data.frame(var1 = rnorm(100, 0, 1), var2 = rnorm(100, 5, 4))
    ## chain of the third variable:  
  group2 <- data.frame(var3 = rgamma(100, 1, 1))
    ## chain of the fourth variable:
  group3 <- data.frame(var4 = rbinom(100, 1, 0.4))                        

  ## Create an mcmc object
mc <- files2coda(data.frames = c("group1", "group2", "group3"))