REQUEST FOR PROPOSAL RFP ISSUE DATE: September 76 th, 7556 TITLE: RFP NUMBER: San José State University Website Redesign F-WR55556686-AL PURCHASING OFFICE CONTACT: DEPARTMENT OFFICIAL: Alex Lebedeff, Mary
CTO on why your agency needs crime analytics software
Predictive Analytics Online Course
In addition, not studied in this analysis is the potential confounding effects of geospatial heterogeneities in crime. For example, changes in local demographics, local intervention measures, etc can lead to spatial heterogeneities [ 99 – 56 ], and these spatial heterogeneities may mask important factors affecting local temporal dynamics [ 57 ]. While this can be studied by examining crime trends in finely granulated geospatial areas, fine granulation necessarily involves arbitrary choices of geospatial boundaries, which, as the granulation is made increasingly finer, increasingly restricts the size of the data sets within each of the areas. This in turn can mask important temporal trends due to the increased relative stochastic fluctuations in the small data sets, and presents serious problems from a predictive analytics perspective. Statistical methods are needed that combine temporal and geospatial information in a scale-invariant manner, while also taking account temporal trends and periodicity in the data (both weekday and annual). The recent work of [ 58 ] introduced a clustering method that takes into account geospatial information, and modeled potential annual periodicity in the data with a first order harmonic. Avenues of future research could potentially involve incorporating the more sophisticated temporal models used in this analysis with scale-invariant geospatial clustering methods, such as the novel Dynamic Covariance Kernel Density Estimation (DCKDE) method previously developed by the authors [ 7 ].