kendall_top: Kendall kernel for top-k rankings

In YunlongJiao/kernrank: Kernel Methods for Rank Data

Description

Calculates Kendall kernel between top-k rankings in time O(klogk), where ties (supposed few) are broken by adopting a convolution kernel averaging compatible rankings without ties.

Usage

kendall_top(x, y)

Arguments

x, y	Vector. If x is numeric, the rank vector converted from x indicate that larger values mean being preferred. NAs replace unobserved values.

Value

Kendall kernel for top-k rankings, where ties (supposed few) are broken by adopting a convolution kernel averaging compatible rankings without ties.

Author(s)

Yunlong Jiao

References

Yunlong Jiao, Jean-Philippe Vert. "The Kendall and Mallows Kernels for Permutations." IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), vol. 40, no. 7, pp. 1755-1769, 2018. DOI:10.1109/TPAMI.2017.2719680

Examples

x <- c(1.5, 0.1, NA, -4, NA)
y <- c(NA, NA, 0, 3, NA)
kendall_top(x, y)

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