Genetic Algorithm for Group Formation

Requirements

• Python 3.x
• matplotlib

Usage

./genetic.py --help

This will display all the relevant arguments for this Python script

To run with the exact arguments from the associated paper (Population Size = 10000, Number of Generation = 5000, Elitism Factor = 3, Number of non-elite individuals to select from parent population = 3, Mutation Chance = 0.5, Number of swaps per mutation = 1, Number of participants = 27, Group size = 3), run

./genetic.py -p 10000 -g 5000 -el 3 -rest 3 -mchance 0.5 -mswaps 1 -n 27 -s 3

You will need a file called rankings.csv with a n x n matrix of preference values starting at the first row and first column. The value in row i and column j will refer to the rating given by the ith participant to the jth participant.

There is some support for running multiple GA instances in parallel. Running:

./genetic-parallel.py

will execute a number of ./genetic.py instances in parallel. This has no command-line arguments - all changes need to be made in the code itself currently. The call to run_genetic(...) and the variable cpus are the main points of modification.