A Python library for voting algorithms.
ballotbox is a Python library that attempts to implement useful voting methodologies.
This project started as a result of conversations and experimentation that took place on the ULS-SIG when discussing self-organizing object meshes.
The current implementation depends upon the following additional software and/or libraries:
For information on the following, please see the sections (or, if reading this file in a working directory, the ./docs subdir):
If you want to develop for ballotbox or use the latest code we’re working on, you can install from the sources. You’ll need bzr installed, and then just do the following:
$ bzr branch lp:ballotbox $ cd ballotbox $ sudo python setup.py install
You can use the setuptools easy_install script to get ballotbox on your system:
$ sudo easy_install ballotbox
You can manually download the source tarball from the Python Package Index by visiting the following URL:
You’ll need to untar and gunzip the source, cd into the source directory, and then you can do the usual:
$ sudo python setup.py install
Checking the Source
Once installed, as long as you have Twisted installed on your system and the trial script in your PATH, you can test the source code by executing this anywhere:
$ trial ballotbox
That will run the test suite and report on the success and failure of any unit tests.
Below is a quick introduction to the usage of the API. The voting method used in this example is simple majority. Simple majority rule voting is mostly useful for binary decisions. Here are some examples involving just two choices:
>>> from ballotbox.ballot import BallotBox >>> from ballotbox.singlewinner.simple import MajorityRuleVoting >>> bb = BallotBox(method=MajorityRuleVoting) >>> bb.batch_votes([("alice", 10000), ("bob", 5000)]) >>> bb.get_winner() [(10000, 'alice')] >>> bb = BallotBox(method=MajorityRuleVoting) >>> bb.batch_votes([("bob", 5000), ("carol", 5001)]) >>> bb.get_winner() [(5001, 'carol')]
This method breaks down with ties and is not guaranteed to work with more than two choices:
>>> bb = BallotBox(method=MajorityRuleVoting) >>> bb.batch_votes([("alice", 8000), ("carol", 8000)]) >>> bb.get_winner() 
No result is returned. Likewise for edge cases with more than two candidates:
>>> bb = BallotBox(method=MajorityRuleVoting) >>> bb.batch_votes([("alice", 5000), ("bob", 4000), ("carol", 3000)]) >>> bb.get_winner() 
[DONE] Convert the criteria classes to use interfaces. Then, instead of subclassing these in the voting methods, they would implement them. This will make zope.interface a dependency.
[DONE] Copy multi-doctest file test suite code from txULS/soom to here.
[DONE] Add a new docs/methods directory and a file for each voting method.
[IN PROGRESS] Finish implementing the single-winner voting methods.
Implement the Multi-winner voting methods.
Implement the proxy and random voting methods.
Do research on distance-based voting:
- in particular, defining a metric space for candidates based on a set of preferences
- then taking the square root of the sum of the squares, getting the n-dimensional distance
[IN PROGRESS] Add doctests for each voting method.
Add a make target that tests the source dist:
- create a temp dir
- untar the dist there
- run the unit tests and doctests against that directory
- Change uppercase doc files to lower case (with the exceptions of the standard ones like README, TODO, etc.)
- Updated the sphinx configuration file to use ballotbox.meta
- Add sidebar links in the generated docs
- Autogenerate ReST files for API docs
- the logic that drives this should go in ballotbox.util.docs
- add a make target that imports and calls the appropriate functions to generate the API docs
- Initial release of ballotbox.