NIJ Real-Time Crime Forecasting

Abstract

Given data from the National Institute of Justice on crime, we will present a model that can predict the likelihood of crime occurrences in certain areas of the city of Portland, Oregon, based upon four different crime categories: all calls-for-service, burglary, street crime, and motor vehicle theft. Using information provided by census tract data from Portland State University, collected over multiple time periods, the goal of this research is to create a model that will be able to accurately forecast hotspots for each of the specified types of crimes in designated areas of Portland, Oregon. This work will present multiple models, using R-Studio, and results in the form of visualizations, probabilities, and maps of the predictions.