Crime Hot Spot Forecasting: Modeling and Comparative Evaluation, Final Project Report

The study drew from research results in the field of forecasting, which have provided a wealth of knowledge on forecast models and experimental designs for comparing and assessing forecast accuracy. The theoretical work which included specification of forecast models, forecast performance measures, and experimental designs for evaluating crime forecasts yielded a crime forecasting system of value to practitioners and researchers. A case study of Pittsburgh, PA, yielded empirical findings. The latter study concluded that for Part I violent crimes, the leading indicator models are far superior to the univariate method for large changes in crime patterns. The leading indicator forecasts are analogous to high-quality leads on locations of significant crime changes and warrant resource allocation. For property crimes, the regression model is significantly better than alternatives for forecasting large crime decreases, but the neural network and univariate methods are best and not significantly different for large increases. Similarly, the neural network and regression models are equally useful for large drug call decreases, but all three methods fail to perform well for large increases. Improvement is needed in leading indicator models for large increases in Part I property crimes and 911 drug calls. Results for all other cases are good for the leading indicator models. Future research that uses additional model components, including fixed effects for demographic and land uses that affect crime levels and additional lag structure for leading indicator variables, will make major improvements in the performance of these models. Based on this research, six recommendations are offered for police, and four recommendations are provided for researchers. 72 references, 13 tables, and 9 figures