mwdbm: Fit a mixture weighted distance-based model

In StatMethRank: Statistical Methods for Ranking Data

Description

This function computes fitting of mixture weighted distance-based model for the given data set of complete rankings.

Usage

mwdbm(dset, G, dset.agg = TRUE, dtype = "Kendall", noise = FALSE,
  iter = 100)

Arguments

dset	data set of complete rankings
G	number of clusters
dset.agg	whether the data set is in the aggregated form (default as FALSE)
dtype	type of the weighted distance measure Kendall or K(default) : "Weighted Kendall's tau", SqrtSpearman or SS : "Square root of weighted Spearman", Spearman or S : "Weighted Spearman", Footrule or F : "Weighted Spearman's footrule"
noise	whether a noise cluster is contained (default as FALSE)
iter	number of iterations of the EM algorithm (default as 100)

Value

a list of the fitting result, containing the following objects: $clusterNum number of clusters (excluding the noise) $dtype type of the distance measure $noise whether a noise cluster is contained $iterNum actual number of iterations of the EM algorithm $convergence whether the complete-data loglikelihood converges $clusterProb probability of each cluster $modalRank modal rankings $weight weight vectors for clusters $trueLoglik the true loglikelihood by the fitted model $squaredPearsonStat the sum of squares of Pearson residuals

Author(s)

Yumin Zhang <[e-mail]>

Examples

data(Croon)
# Time comu
# mwdbm(Croon, 3)

StatMethRank documentation built on Jan. 15, 2017, 8:59 p.m.