pdist

Ruby gem to compare permutations using distance measures/metrics.

Install and pre-requisites

gem install pdist

OR download here

You must also have the diff-lcs gem installed, to use the longest common subsequence method (see below).

Usage

Each method of the PDist class takes as input, two arrays of the same unique objects (i.e. both are the same array ordered differently).

The two input arrays are permutations of eachother, and the chosen method calculates the 'distance' between these permutations, according to a certain measure/metric.

For example take these two permutations of the integers 1..5:

a = [1,2,3,4,5]
b = [2,1,3,5,4]

Each of the methods below gives as output a float of value x, where 0.0 <= x <= 1.0

This output value is the distance between the permutations as attributed by the metric, normalized in the range 0.0 to 1.0

Output values closer to 0.0 indicate closely matching permutations, where 0.0 is a perfect match, and 1.0 is the maximum possible distance.

To get the un-normalized distance for each of these metrics, simply append _raw to any of the methods below (before the arguments)

Deviation distance

PDist.deviation(a,b)

### Square deviation distance

PDist.square(a,b)

Hamming distance

PDist.hamming(a,b)

The R distance (reversed)

PDist.rdist(a,b)

The longest common sub-sequence

PDist.lcs(a,b)

Kendall's tau

PDist.kendalls_tau(a,b)

Mathematical descriptions of methods

Each of the methods in the PDist class is described mathematically by Sevaux and Kenneth (2005)

The authors also describe several other distance metrics for permutation comparison. Forkers who wish to implement these methods or others are most welcome.