Gerrymandering_Detection

Gerrymandering is an issue of paramount importance in several parts of the world, particularly in the US, and has been the subject not only of a significant amount of literature, but also of recent Supreme Court cases and official redistrict plans. It poses a great threat to democracy, and the need for the development of appropriate detecting and preventing measures is evident. In this paper, we employ machine learning as a predicting measure for gerrymandering. In particular, via combining appropriate image district data and their associated demographic data, we train deep learning models that are able to detect gerrymandered districts with rather high accuracy. Our results show for the first time the potential effectiveness of deep learning as a tool against gerrymandering, and opens up avenues for further investigations.