tsTransform {genMCMCDiag}

Transforms a list of MCMC chains into a list of data.frames using the TS transformation

Description

Transforms a list of MCMC chains into a list of data.frames using the TS transformation

Usage

tsTransform(
  mhDraws,
  distance,
  minDist = 0,
  fuzzy = FALSE,
  fuzzyDist = 1,
  verbose = FALSE,
  ...
)

Arguments

mhDraws

List. Each element is a single chain from an MCMC algorithm. Each element should be a numeric vector (for univariate draws), or a list.

distance

Distance function defined on the space of MCMC draws. Should operate pairwise on the elements of the given chains. See details.

minDist

Numeric. Value which specifies the minimum possible distance for two draws which are not equal. See tsTransform details.

fuzzy

Logical. If TRUE computes an approximate version of the TS algorithm. See tsTransform details.

fuzzyDist

Numeric. Parameter for approximate version of ts algorithm. See tsTransform details.

verbose

Logical. If TRUE, function prints out information about approximate computation time

...

Catches extra arguments. Not used.

Details

The TS transformation sets up a traveling salesman algorithm by calculating the pair-wise distances between each unique draw from the mhDraws and solving the resulting TS problem with the nearest neighbor (NN) algorithm.

minDist can be used to speed up the algorithm if it is known that when x != y then distance(x, y) >= minDist. Otherwise this should be ignored.

The fuzzy approximation of the algorithm works by splitting the unique draws into smaller sets each containing at most 1% of all unique draws, and fitting the NN algorithm within each set, and then on the resulting 'end points' of each set. The sets are created by randomly selecting a representative draw and then putting the 'closest' draws with distance less than fuzzyDist into that set, until the set contains 1% of all unique draws. The fuzzy approximation can GREATLY reduce computation time, unless the fuzzyDistance specified is too small.

Value

List of data.frames with columns 'val' which is the TS transformation of each MCMC draw, and 't' which gives the within-chain ordering of the MCMC draws. Each data.frame is a separate chain.