Can we identify factors that make certain cities at a higher risk of suffering police shooting catastrophes? Building confidence around the answer to this question could have an impact on public policy, police training, education and beyond.
We use a Quantum SVM algorithm to run a classification model predicting whether or not a city will have "high" police killings or "low" police killings. Due to computation constraints, we trained on a limited number of observations and predictors, but we still obtained a test accuracy of 0.85.
Studying events of police shootings ex-post creates the risk of making claims based on spurious relationships. Human behavior can be illogical, so even in a world of perfect data availability, unavoidable catastrophes will occur
Our initial data exploration and analysis uncovered mental-health, socio-economic and racial phenomena that have showed correlations to police shootings. These are important social issues that should be addressed regardless of model findings.
- @Sara-ShiHo
- @TiffanyKe888
- @isaacNorthwestern
- @lcgenereux13
- @matthewko1698
-
Kaggle data: Fatal Police Shootings in the US and the original source, WaPo data on police shootings
-
Police agency-level data from Campaign Zero Nationwide Police Scorecard Data
-
Educational attainment for the U.S., States, and counties, 1970-2019, Economic Research Service of the U.S. Department of Agriculture