mdir (version 0.9.0)

predictFromMultipleChains: Predict from multiple MCMC chains

Description

Applies a burn in to and finds a point estimate by combining multiple chains of ``callMDI``.

Usage

predictFromMultipleChains(
  mcmc_outputs,
  burn,
  point_estimate_method = "median",
  construct_psm = FALSE,
  chains_already_processed = FALSE
)

Value

A named list of quantities related to prediction/clustering:

* ``allocation_probability``: List with an $(N x K)$ matrix for each semi-supervised view. The point estimate of the allocation probabilities for each data point to each class.

* ``prob``: List with an $N$ vector for each semi-supervised view. The point estimate of the probability of being allocated to the class with the highest probability.

* ``pred``: List of $N$ vectorsfor each semi-supervised view. The predicted class for each sample.

* ``allocations``: List of sampled allocations for each view. Columns correspond to items being clustered, rows to MCMC samples.

Arguments

mcmc_outputs

Output from ``runMCMCChains``

burn

The number of MCMC samples to drop as part of a burn in.

point_estimate_method

Summary statistic used to define the point estimate. Must be ``'mean'`` or ``'median'``. ``'median'`` is the default.

construct_psm

Logical indicating if PSMs be constructed in the unsupervised views. Defaults to FALSE. If TRUE the PSM is constructed and this is used to infer the point estimate rather than the sampled partitions.

chains_already_processed

Logical indicating if the the chains have been processed already.

Examples

# \donttest{
N <- 100
X <- matrix(c(rnorm(N, 0, 1), rnorm(N, 3, 1)), ncol = 2, byrow = TRUE)
Y <- matrix(c(rnorm(N, 0, 1), rnorm(N, 3, 1)), ncol = 2, byrow = TRUE)

truth <- c(rep(1, N / 2), rep(2, N / 2))
data_modelled <- list(X, Y)

V <- length(data_modelled)

# MCMC parameters
R <- 5000
thin <- 50
burn <- 1000

K_max <- 10
K <- rep(K_max, V)
types <- rep("G", V)

n_chains <- 3
mcmc_out <- runMCMCChains(data_modelled, n_chains, R, thin, types, K = K)
predictFromMultipleChains(mcmc_out, burn)
# }