ron-rivest/2017-kenya-audit

Code and documentation related to possible audit of 2017 election re-run in Kenya.

2017-kenya-audit's Issues

Add unit tests for functions in the kenya.py

We should add unit tests in the tests folder for the kenya.py code

Allow multiple audit results for same polling station.

This will be used when multiple copies of a 34A form are reviewed by external observers.
The system should be modified to choose one audit interpretation from the list of interpretations instead of using the only interpretation available.

Add Unit Tests for Opt.py

We should add unit tests for the Opt.py file into the tests folder.

Create additional malicious actor models to run simulations.

We should create additional Python scripts to generate audit.csv files based off of sites.csv files. Using the bias_results in pull request #6: https://github.com/zperumal/2017-kenya-audit/blob/69e718eda5239f5fe409d2110e8f458d0641dc46/data/bias_results.py as a basis. Bias_results.py takes a candidate as input and changes the audit results for that candidate by multiplying it by a float bias.

Some different attacker models of manipulating results could include:

1. Modify one candidate by specific specified amount
2. Modify losing candidates randomly
3. Modify one candidate in specific county or polling station
4. Modify both winner and loser in specified counties/locations