Modeling police-citizen relations using Network Analysis
The city of Chicago recently published a large dataset of police complaints filed by citizens. Each report contains a text explaining the complaint as well as a list of the police the complaint is against. This lends itself to the creation of a co-offender network - a graph where each node is a policeman/woman and nodes are connected if they are listed on more than some threshold of complaints together. We then use diffusion modeling to investigate if there are certain nodes which are 'infecting' other nodes with bad behavior. I joined this project in November 2016.