Predictive Spatial Models for Crime Analysis
You can incorporate predictive statistical models into crime analysis methods to produce analytics that describe and predict where and what kinds of criminal activity are likely to occur.
Predictive spatial models can help you predict the behavior, location, or criminal activities of repeat offenders. You can also apply statistical methods to spatio-temporal data to ascertain causative or correlative variables relevant to crime and law enforcement.
The following list includes types of approaches that are helpful in spatial predictive modeling for crime analysis:
- Clustering: You can use kernel density estimation methods to quantify the spatial density of criminal activities and to generate comparative measures between the densities of criminal activity relative to the base population of the affected area.
Kernel density estimation (KDE) is a smoothing method that works by placing a kernel — or, a weighting function that is useful for quantifying density — on each data point in the dataset and then summing the kernels to generate a kernel density estimate for the overall region.
- Advanced spatial statistics: One example of this is to use regression analysis to establish how one or more independent crime variables directly cause, or correlate with, a dependent crime variable. Lastly, advanced spatial statistics are used to make behavioral predictions for repeat offenders and to predict future criminal activity based on historical records on criminal behavior and information about present conditions.