See the COMSOC community page for an overview of other software tools related to Computational Social Choice.
The package can be installed using the pip3 package manager:
pip3 install pref_votingNote: If you have both Python 2 and Python 3 installed on your system, make sure to use pip3 instead of pip to install packages for Python 3. Alternatively, you can use python3 -m pip to ensure you're using the correct version of pip. If you have modified your system's defaults or soft links, adjust accordingly.
See the installation guide for more detailed instructions.
A profile (of linear orders over the candidates) is created by initializing a Profile class object. Simply provide a list of rankings (each ranking is a tuple of numbers) and a list giving the number of voters with each ranking:
from pref_voting.profiles import Profile
rankings = [
(0, 1, 2, 3), # candidate 0 is ranked first, candidate 1 is ranked second, candidate 2 is ranked 3rd, and candidate 3 is ranked last.
(2, 3, 1, 0),
(3, 1, 2, 0),
(1, 2, 0, 3),
(1, 3, 2, 0)]
rcounts = [5, 3, 2, 4, 3] # 5 voters submitted the first ranking (0, 1, 2, 3), 3 voters submitted the second ranking, and so on.
prof = Profile(rankings, rcounts=rcounts)
prof.display() # display the profileThe function generate_profile is used to generate a profile for a given number of candidates and voters:
from pref_voting.generate_profiles import generate_profile
# generate a profile using the Impartial Culture probability model
prof = generate_profile(3, 4) # prof is a Profile object with 3 candidates and 4 voters
# generate a profile using the Impartial Anonymous Culture probability model
prof = generate_profile(3, 4, probmod = "IAC") # prof is a Profile object with 3 candidates and 4 voters The Profile class has a number of methods that can be used to analyze the profile. For example, to determine the margin of victory between two candidates, the plurality scores, the Copeland scores, the Borda scores, the Condorcet winner, the weak Condorcet winner, and the Condorcet loser, and whether the profile is uniquely weighted, use the following code:
prof = Profile([
[2, 1, 0, 3],
[3, 2, 0, 1],
[3, 1, 0, 2]],
rcounts=[2, 2, 3])
prof.display()
print(f"The margin of 1 over 3 is {prof.margin(1, 3)}")
print(f"The Plurality scores are {prof.plurality_scores()}")
print(f"The Copeland scores are {prof.copeland_scores()}")
print(f"The Borda scores are {prof.borda_scores()}")
print(f"The Condorcet winner is {prof.condorcet_winner()}")
print(f"The weak Condorcet winner is {prof.weak_condorcet_winner()}")
print(f"The Condorcet loser is {prof.condorcet_loser()}")
print(f"The profile is uniquely weighted: {prof.is_uniquely_weighted()}")To use one of the many voting methods, import the function from pref_voting.voting_methods and apply it to the profile:
from pref_voting.generate_profiles import generate_profile
from pref_voting.voting_methods import *
prof = generate_profile(3, 4) # create a profile with 3 candidates and 4 voters
split_cycle(prof) # returns the sorted list of winning candidates
split_cycle.display(prof) # displays the winning candidatesAdditional notebooks that demonstrate how to use the package can be found in the examples directory
Some interesting political elections are analyzed using pref_voting in the election-analysis repository.
Consult the documentation https://pref-voting.readthedocs.io for a complete overview of the package.
To ensure that the package is working correctly, you can run the test suite using pytest. The test files are located in the tests directory. Follow the instructions below based on your setup.
- Python 3.9 or higher: Ensure you have a compatible version of Python installed.
pytest: Installpytestif it's not already installed.
If you are using Poetry to manage your dependencies, run the tests with:
poetry run pytest
From the command line, run:
pytestFor more detailed output, add the -v or --verbose flag:
pytest -vIf you would like to contribute to the project, please see the contributing guidelines.
Feel free to send an email if you have questions about the project.