rbo: Rank biased overlap (Webber et al., 2010)

In cbg-ethz/gespeR: Gene-Specific Phenotype EstimatoR

Description

Evaluates the rank biased overlap (rbo) of two ranked lists based on formula based on (32) from "A Similarity Measure for Indefinite Rankings" (Webber et al.). Two ranked lists with high rbo are very similar, wheras low rbo indicates dissimilar lists. rbo ranges between 0 and 1. In this method the extrapolated version of rbo is implemented.

Usage

rbo(s, t, p, k = floor(max(length(s), length(t))/2), side = c("top",
  "bottom"), mid = NULL, uneven.lengths = TRUE)

Arguments

s	List 1
t	List 2
p	Weighting parameter in [0, 1]. High p implies strong emphasis on top ranked elements
k	Evaluation depth for extrapolation
side	Evaluate similarity between the top or the bottom of the ranked lists
mid	Set the mid point to for example only consider positive or negative scores
uneven.lengths	Indicator if lists have uneven lengths

Value

rank biased overlap (rbo)

Author(s)

Fabian Schmich

See Also

concordance

Examples

a <- rnorm(26)
b <- rnorm(26)
names(a) <- names(b) <- LETTERS
rbo(a, b, p = 0.95)

cbg-ethz/gespeR documentation built on May 13, 2019, 2:02 p.m.