Forecasting Crime Rates Utilizing Machine Learning and Economic Indicators

Abstract: The following project utilized the correlation between economic indicators and crime rates to build tree-based machine learning algorithms that predict monthly crime rates in Ohio. The goal of the study was to answer how economic variables could be used in forecasting county level crime rates. A sample of 9 counties were utilized in building regression trees, random forests and evolutionary random forest models. The study concluded that lagged crime rates provided the best predictive ability in forecasting models, while economic variables provided more accuracy to property crime. The study could be furthered by creating models at the department level and expanding the amount of data utilized.