Vote-processing rules for combining control recommendations from multiple models

Vote-processing rules to accompany the publication entitled Vote-processing rules for combining rankings of control interventions from multiple models. This repo includes code for implementing the vote-processing rules, functions for translating model output into "votes", and several tests for all functions in this repository.

The repository for implementing the analysis that is presented in the manuscript above is found at voting_systems_epi_analysis. It depends on having this module installed as it uses the functions defined herein.

Usage

This module was built using Python version 3.8. The file requirements.txt provides a list of Python modules upon which this module depends and a virtual environment can be set up, activated, and the requirements for this module installed in the following manner:

python -m venv venv
source venv/bin/activate
pip install -r voting_systems/tests/requirements.txt

Deactivate the environment using deactivate.

Example usage

Consider a group of individuals ranking their favourite planets amongst the three closest to the sun (i.e. Mercury, Venus, Earth).

import voting_systems as voting

# Define an example set of ballots; rows are 'voters'
# order is the preference; for instance, the first
# voter thought Mercury is the best, followed by Earth
# then Venus
example_planets = np.array([
    ["Mercury", "Earth",   "Venus"],
    ["Mercury", "Earth",   "Venus"],
    ["Mercury", "Venus",   "Earth"],
    ["Earth",   "Venus",   "Mercury"],
    ["Earth",   "Venus",   "Mercury"],
    ["Earth",   "Venus",   "Mercury"],
    ["Venus",   "Mercury", "Earth"]
])

# Process the ballots using Coombs method
(winner, winner_index), (candidates, removed) = voting.coombs_method(example_planets)
print(winner)
> "Earth"

# Process the ballots using Alternative vote method
(winner, winner_index), (candidates, removed) = voting.coombs_method(example_planets)
print(winner)
> "Mercury"

Tests

All vote processing-rules are tested using pytest. Tests can be run in the following manner if called from this folder (after creating and activating a virtual environment):

pytest

All tests are in test_voting_rules.py.

Publication

The publication associated with this code is the following:

• Probert et al., (2022) Vote-processing rules for combining control recommendations from multiple models. Phil. Trans. R. Soc. A. 380: 20210314. https://doi.org/10.1098/rsta.2021.0314.