Moving from reactive to proactive policing
According to the data released by FBI, there is an increase in violent crimes by 3.9% and property-related crimes by 2.6% over the last year. In spite of all the necessary measures in place, the crime rates are only increasing. In such a scenario, proactively preventing crime is a better approach than post-crime investigation. Analytics- and machine learning-based “crime prediction” methodologies can not only save millions by preventing robberies, burglaries etc., but also aid in better utilization of limited police resources.
As per studies conducted by the University of California, crime in any area follows the same pattern as earthquake aftershocks. It is difficult to predict an earthquake, but once it happens the aftershocks following it are quite predictable. Same is true for the crimes happening in a geographical area. And, once successful, criminals tend to operate in similar conditions. Combined analysis of past crime data and the other influential parameters can predict location, time and category of crime. Methodologies like time series analysis, gradient boosting machine, random forest, etc., give prediction accuracy as high as 60-74%.
Predicting crime by applying analytics on data feeds from various sources
In order to learn more about this, please download the insight presentation.
Avaiarasi S, Director – Delivery (IoT)
Kamakya C, Project Manager – IoT
Sarvagya Nayak, Business Analyst – Insights