mmMCMCFit: Fit Mixed Membership models using a Metropolis Hastings...
In ysamwang/mixedMem: Tools for Discrete Multivariate Mixed Membership Models

Description Usage Arguments Details See Also

Takes samples form the posterior distribution of a mixed membership model using a Metropolis-Hastings within Gibbs sampler.

mmMCMCFit(model, burnIn = 20000, samples = 1000, thin = 10, print = 100,
  fileNames = c("theta.csv", "alpha.csv", "ksi.csv", "lambda.csv", "z.csv",
  "p.csv", "rho.csv"), newFiles = 1, omega = 100, eta = 1,
  whichWrite = c(1, 1, 1, 0, 0, 1, 0))

`model`	a `mixedMemModelMCMC` object created by the `mixedMemModelMCMC` constructor
`burnIn`	non-negative integer indicating the number of burn in samples before recording the first sample
`samples`	positive integer indicating the number of samples to record
`thin`	positive integer indicating how to thin the samples
`print`	positive integer indicating how often to print an update to the R console
`fileNames`	list of files locations to write samples. Should be vector of strings with length 5 (7 if using the extended mixed membership model) corresponding to: Theta, Alpha_0, Ksi, Lambda, Z, (P, rho).
`newFiles`	0 if samples should be appended to existing files; 1 if samples should overwrite any existing files
`omega`	tuning parameter for MH step for alpha.
`eta`	tuning parameter for MH step for ksi
`whichWrite`	which parameters to write to disk. Writing the individual parameters can significantly increase computational time when the number of individuals is large
`extended`	boolean whether to estimate the extended model or not

mmMCMCFit draws samples from the posterior distribution of a mixed membership model given a set of observed data. The sampler takes Metropolis Hastings steps to sample the α_0 and ξ parameters and samples the remaining θ, λ_i and Z parameters with a Gibbs sampler.

mixedMemModelMCMC

ysamwang/mixedMem documentation built on May 4, 2019, 5:33 p.m.