Abstract
Safety plays a crucial role in providing a conducive environment for education. Predictive policing is one such area of study where statistical prediction is conducted to forecast the probability of a crime occurrence in the future and can help promote proactive policing to reduce the overall crime rate. The result of preliminary research has shown that predictive policing can help predict future crime for the CSUS campus with above 70% accuracy. However, such technology has not been implemented into the application level to benefit all CSUS community members yet. This project aims at designing and developing an application which implements the predictive policing technology, so that CSUS community members can interact with and benefit from it. Service-Oriented-Architecture (SOA) has been adopted to keep the architecture both scalable and maintainable. It can also enable modular design so that different modules can work with each other independently and each module can be replaced and updated in isolation. The crime log of the past 7 years from the CSUS police department is used as raw data. For prediction of future crime, TensorFlow along with data analysis Python libraries is used. The core service layer is built using Flask microframework in together with Python libraries for date time and mapping features. The user interface is constructed as a JavaScript-HTML webpage rendering the parameters as per Jinja template format. The application has been deployed to Amazon Web Services (AWS) cloud and accessible via a public IP address. The application will enable users to identify locations, crime types, and date of occurrence of potential crime within the campus, understand crime distribution, and also report a crime. It can also help the CSUS police department plan more strategically and disseminate information more easily. Moreover, this project itself can serve as a pilot project which can help deploy predictive analytic technologies into more services/applications onto the CSUS campus.