In a working paper presented to UNC Charlotte’s School of Data Science, Dr. Radford evaluates the predictive power of prior protest events to forecast subsequent policing actions. Below are videos of predicted probabilities.

The Poster

SDS Poster 2024

Full size image.

Police Presence Predictions

  • Upper left: actual observations of police presence.
  • Upper right: predictions made with only structural predictors.
  • Lower left: predictions made with structural predictors, weather conditions, and time.
  • Lower right: predictions made with structural predictors, weather conditions, time, and prior protest-related events.

Police Vehicle Predictions

  • Upper left: actual observations of police vehicles.
  • Upper right: predictions made with only structural predictors.
  • Lower left: predictions made with structural predictors, weather conditions, and time.
  • Lower right: predictions made with structural predictors, weather conditions, time, and prior protest-related events.

PANDAS Researchers


Benjamin J. Radford. Assistant Professor, UNC Charlotte. [Homepage] [Github]

Learn more about the PANDAS Lab Team.