bPlackettLuceModel: Bayesian Plackett-Luce model for ranking analysis

In b0rxa/scmamp: Statistical Comparison of Multiple Algorithms in Multiple Problems

Bayesian Plackett-Luce model for ranking analysis

Description

Bayesian model based on the Plackett-Luce distribution over rankings to analyse multiple algorithms in multiple problems

Usage

bPlackettLuceModel(
  x.matrix,
  min = TRUE,
  prior = rep(1, ncol(x.matrix)),
  nsim = 2000,
  nchains = 8,
  parallel = TRUE,
  stan.output.file = NULL,
  seed = as.numeric(Sys.time()),
  ...
)

Arguments

x.matrix	Sample of performance of the algorithms. Each column is an algorithm and each row is a problem
min	Logical value indicating which values should be ranked in first postion. If TRUE, the smallest value will be the first ranked and if FALSE the first will be the one with the highest value
prior	Hyperparameters of the prior distribution of the weights. It should be a vector of size equal to the number of algorithms of real valued numbers greater than 0By default, all equal to 1
parallel	Logical value. If true, Stan code is executed in parallel
stan.output.file	String containing the base name for the output files produced by Stan. If NULL, no files are stored.
seed	Optional parameter used to fix the random seed
...	Additional arguments for the rstan::stan function that runs the analysis
nchain	Number of MC chains to be simulated. As half the simulations are used for the warm-up, the total number of simulations will be nchain*nsim/2

Value

A list with the following elements:

method	a string with the name of the method used
parameters	parameters used by the method
posterior.weights	a vector with the weights sampled from the posterior distribution
expected.win.prob	for each algorithm, the expected posterior probability of being the best
expceted.mode.rank	for each algorithm the expected rank in the most probable ranking
additional	complete results produced by the Stan program

Examples

n.alg  <- 5
n.inst <- 25
x.matrix <- matrix(runif(n.alg*n.inst), ncol=n.alg)
colnames(x.matrix) <- paste("Alg", 1:n.alg, sep="")
rownames(x.matrix) <- paste("Inst", 1:n.inst, sep="")

res <- bPlackettLuceModel(x.matrix, min=FALSE, nsim=2000, nchains=3)

res$expected.win.prob
res$expected.mode.rank

b0rxa/scmamp documentation built on Jan. 17, 2024, 10:49 a.m.