PredPol: Predictive analytics to place police officers at the right time and place to increase chances of preventing crime.
PredPol is an analytic platform that assign crime probabilities to different places and times. With these insights, Police Departments can assign and schedule their patrols and officers to these dynamic “hot spots”, preventing crime and reducing significant costs to taxpayers.
This platform was created by a combined team of PhD mathematicians and social scientists at Santa Clara University and University of California Los Angeles. For more than six six years they worked in lose relationship and collaboration from crime analysts and officers at the Santa Cruz and Los Angeles Police Departments in California.
Overall, PredPol is an analytic platform that assign crime probabilities to different places and times. With these insights, Police Departments can assign and schedule their patrols and officers to these dynamic “hot spots”, preventing crime, reducing significant costs to taxpayers and engaging more effectively with the community.
PredPol basically takes three different data points from past criminality: type (for example gun crime, residential burglary, etc.), place and time. A proprietary algorithm takes this input and creates predictions for times and places that different crimes are most likely to happen. An important characteristic is that this technology does not replace the intuition and experience of the officer, but rather provide a value added layer to use their patrol time in a more efficient way.
Value creation and capture:
- Reduction in crime rates, by simply arrive to “hot spots” before the crime is likely to occur. PredPol informed that during the initial pilot launch in Atlanta in 2013, aggregate crime decreased by 8% and 9% in the two areas that first activated PredPol usage. Of the four zones where PredPol was not activated, actually crime increased or remained flat. This Police Department was so motivated with this results that decided to implement PredPol across all the city, which has witness a relevant 19% drop their overall crime rate.
- Cost reduction: Reducing crime also creates a very important cost reduction for the taxpayers. In fact, the fewer the number of crimes, the less will be the activity of law enforcement agencies, jails and courts.
- Community Engagement: Given that valuable resources (time and money) are saved from the decrease in crime rates, Police Departments and Law Enforcement Agencies can work even higher with local communities, not only preventing long-term crime, but also supporting people in other aspects.
PredPol is able to capture part of the value that creates by using a subscription based model with different Police Departments. This institution will pay once a year a cost that is based on population size, rather than number of unique users. Also, no new hires nor capex is required by Police Departments, which can receive unlimited platform updates.
Of course that there are a lot of factors that can will have an impact in the behavior of crime rates, and probably the decline observed can be attributed in some extend to PredPol. However, we have to be very cautious about making this conclusion. In fact, some Police Departments (e.g. Hagerstown, Maryland) have tried to “prove” that the decline cannot be attributed to PredPol, and so far, they have not yet found any evidence to support it. Because of this, its very likely that PredPol can take credits for it!
Student comments on PredPol: Predictive analytics to place police officers at the right time and place to increase chances of preventing crime.
This is fascinating. You’d think cities in general would be happy with deployment of law enforcement based on crime rates, but I’ve seen communities where crime is not that probable complain if they think their communities are being ignored, or if petty theft is being tolerated while police go after trouble spots. I’d be interested to see how the public responds to a program like this. In any case: very cool idea.
Interesting post and idea. Definitely agree on the difficulty in attributing responsibility / causation. I could also see police departments using this service for a year and then perhaps figuring out that certain neighborhoods or areas are more prone for certain types of crimes and once again start to rely on their own intuition.