A single link to the first track to allow the export script to build the search page
  • Undergraduate Poster Abstracts
  • SAT-828 A MATHEMATICAL MODEL FOR SETTING CRIME REDUCTION TARGETS

    • Anthony Gusman ;

    SAT-828

    A MATHEMATICAL MODEL FOR SETTING CRIME REDUCTION TARGETS

    Anthony Gusman1, Erik Bates2, Stephanie Sanchez3, Sarah Verros4, Yoon-Sik Cho5.

    1Vanguard University, Costa Mesa, CA, 2Michigan State University, East Lansing, MI, 3University of California, Los Angeles, Los Angeles, CA, 4Colorado School of Mines, Golden, CO 5Yoon-Sik Cho, University of Southern California, Los Angeles, CA.

    Traditionally, the Los Angeles Police Department (LAPD) sets crime reduction goals for each division according to a linear model of 5% reduction from the previous year's levels. Such an approach may not account for seasonal patterns, random environmental fluctuations, limiting threshold conditions, nor the unique spatial context of each policing area. These limitations may result in potentially inaccurate performance assessments. We seek to develop a generalizable approach for setting reasonable crime reduction goals that accounts for each of these factors. The primary challenges are noise in the data that makes the extraction of patterns more difficult and the inherent uncertainty of forecasting crime rates over long-term time scales (months to years). The LAPD has agreed to supply appropriate data. Data smoothing over mid- and long-term windows will be used to explore general behavior and to identify potential seasonal trends, which will then be extrapolated via pattern-detection forecasting methods. To test our forecasting procedure, half of the data will be used as a training set, and accuracy will be quantified using standard goodness-of-fit tests. Furthermore, we will measure the variance due to environmental noise by modeling crime levels as the intensity of a Gaussian Cox process. These techniques will guide the setting of crime reduction goals for policing areas. When compared to the traditional method of setting crime reduction targets, we anticipate that our method will yield more realistic and accurate crime reduction goals. Moreover, we expect that our approach could be extended to other agencies, offering a data-informed alternative to arbitrary crime reduction target setting.