RankAggreg: Common ranking aggregation methods for permutations

In YunlongJiao/kernrank: Kernel Methods for Rank Data

Description

Used to update modal sequences of each cluster in the EM algorithm when fitting Mallows mixture models.

Usage

RankAggreg(r, z = NULL, infos = NULL, perm = NULL, key = c("borda",
  "copeland", "brute"))

Arguments

r	A vector or a matrix of sequences in rows.
z	A vector of weights/frequencies of observations or a matrix of probability of cluster membership for each sequence and each cluster. Set by default a constant vector of 1.
infos	The result of calling KendallInfo. Optional for speeding up computation.
perm	A matrix of full permutations for brute-force search of optimal Kemeny consensus. Only effective when "key" equals to "brute".
key	A character string indicating the ranking aggregation method to find centers. borda denotes the Borda count copeland denotes the Copeland's aggregated ranking brute denotes the optimal Kemeny consensus found by brute-force search

Value

List of length 1 if "z" is a vector, or ncol(z) if "z" is a matrix, each entry being the modal sequence in each cluster.

Author(s)

Yunlong Jiao

See Also

KendallInfo

Examples

r <- do.call("rbind", list(1:5, 5:1, c(2,4,1,5,3)))
RankAggreg(r, key = "borda") # Borda count for sequences in "r"

YunlongJiao/kernrank documentation built on May 10, 2019, 1:13 a.m.