A Theory-Driven Algorithm for Real-Time Crime Hot Spot Forecasting, Report to NIJ

• NCJ Number: 251179
• Author(s): Yong Jei Lee; Soo Hyun O; John E. Eck
• Date Published: October 2017
• Length: 9 pages
• Grant Number(s): 2016-NIJ-Challenge-0017
• Agencies: NIJ-Sponsored
• Publication Type: Report (Technical); Report (Technical Assistance); Program/Project Description

Annotation

Since there is a high volume of crime hot spot misclassifications and a lack of theoretical support for existing forecasting algorithms, this paper fills these gaps by suggesting a new algorithm by operationalizing two different theories.

Abstract

First, a population heterogeneity framework is used to find places that are consistently experiencing crimes in the forecasted month. Second, in order to assess the elevated risk at place, the new algorithm uses state dependence model of the number of crimes in the time period prior to the forecasted month. This theory-driven algorithm is implemented in Microsoft-Excel, making it transparent and simple to apply. Experiments have shown high accuracy and high efficiency in hot spot forecasting. The results also show how basic theories could lead researchers to create a sound algorithm for hot spot forecasting. Future research will focus on improving the performance of the forecasting algorithm by incorporating other features of place. There will also be an effort to forecast hot spots within a time frame shorter than 1 month. 1 figure, 1 table, 12 references, and appended descriptive statistics of all types of calls for service (CFS)