Free Newsletter
Register for our Free Newsletters
Newsletter
Zones
Airport, Border and Port Safety and Security
Left Nav Sep
Bioterrorism
Left Nav Sep
Civil Aircraft
Left Nav Sep
Civil Airlines, Airports and Services
Left Nav Sep
Communications and Navigation
Left Nav Sep
Education, Training and Professional Services
Left Nav Sep
Manufacturing and Materials
Left Nav Sep
Military and Defence Facilities
Left Nav Sep
Military Aviation
Left Nav Sep
Military Vehicles
Left Nav Sep
Naval Systems
Left Nav Sep
Personal Equipment
Left Nav Sep
Software and IT Services
Left Nav Sep
Space and Satellite
Left Nav Sep
Weapons, Ammunition and Explosives
Left Nav Sep
View All
Other Carouselweb publications
Carousel Web
Defense File Logo
New Material Logo
Pro Health Service Zone
Pro Manufacturing Zone
Prosecurity Zone
Web Lec Logo
Pro Engineering Zone
 
 
 
News

Using predictive policing for improved effectiveness

Intelligent Software Solutions : 20 October, 2015  (Special Report)
Guy Pascarella, Director of Engineering, Federal Systems at Intelligent Software Solutions explains how predictive policing tools have been adapted to incorporate experience and environment into everyday policing practices

"A man shuffles along the pavement shifting his jacket against the cold. He’s been out of work for months and he needs money. Sliding his hand into his pocket, he hopes to find some crisp banknotes but all he feels is the cold steel of his 9mm pistol. A game between two popular hockey teams at the large city centre arena will be ending soon and the surrounding streets and underground stations will be filled with fans, many of whom won’t be street smart enough to guard against experienced pickpockets or recognise an ideal situation for an mugging. Clouds and a sliver of moon make for a dark night and dark alleys. This will be easy. A distraction here, an intimidating flash of his handgun there and he has what he wants. He’ll step onto a crowded bus or underground railway carriage and will be a long way away before the police are even alerted. He rounds the corner some distance from the arena only to discover a dozen cycle police scanning the area and patrolling the streets and alleys. The man curses under his breath and turns on his heels. Tonight is not the night for this."

Predictive Policing is a policing methodology where crimes are deterred or prevented. It is not the ability to predict where and when a specific crime will happen, since that only happens in popular sci-fi films. However, predictive policing tools will indicate where and when crimes are most likely to happen. This is incredibly helpful in today’s public safety operations. Budgets and forces are shrinking, which means decisions need to be deliberate with an eye towards maximising everyone’s usefulness while still keeping peace and reducing crime rates.

Two major factors that are not prevalent within current predictive policing systems are officer experience and environmental conditions. Officers and agents have built large amounts of personal knowledge about the areas and environment that they work within that cannot be located within traditional data repositories.  This information exists only within their minds and includes traffic and commuting information, weather, major event schedules, mass transit schedules, air quality, social media trends and other environmental conditions which are sometimes considered individually within predictive policing systems but not as a whole and not in combination with historical crime information and the experience of individual officers.  

Seasoned law enforcement personnel should be able to tune a system to look for patterns of data based on their experiences.  Providing officers with tools to capture their experiences within the relevant environment and then be alerted when patterns occur results in a proactive approach instead of a reactive one, providing a practical element to everyday policing.  Being able to use the experiences recorded by the human brain and incorporating environmental conditions within a multisource data and analytic system can enhance the ability for the officer to be at the right place at the right time. As data quantity and complexity increases and resources decrease, officers will need systems that enable resources to be used in the most efficient, effective ways possible.

Statistics

There are many ways to accomplish predictive policing but in the realms of automated tools, there are 2 basic evolutions and the future where these technologies are heading.  The first generation uses crime statistics to give a picture of where crimes are most likely to happen because they always happen at these locations.  This gives a relatively low level of fidelity and would focus forces on only a few hot areas, which deprives other locations of the benefits of police protection.

Algorithms

The second generation takes this further by applying research into crimes to these statistics.  For example, applying Near Repeat Phenomenon will generate hotspots around specific crimes after they happen.  Another research topic typically employed is Risk Terrain Modeling, which dictates the likelihood of a crime happening near an off-licence may be greater than a crime happening around a school.  These are more reactive than proactive techniques since they require a crime to be committed before they can tell you where future crimes are likely to occur.

Modeling and Intelligence

The use of raw statistics and more refined algorithms has served the community well so far but criminals can easily adapt to thwart these systems since they simply react to crimes that have already happened.  The third generation of behaviour-based and predictive analysis tools intend to overcome these limitations by extending second generation tools in three ways.  

First, it includes environmental information with the crime data such as forecasted temperatures, weather and moon phases, major events such as concerts and festivals, mass transit information and economic indicators.  

Secondly, information about the area under watch is collected and used to drive the prediction engine.  

Finally, it doesn’t use just simple statistical algorithms like k-means clustering but introduces the use of behaviour models.  This gives officers and subject matter experts the ability to capture their personal knowledge and experience and have a prediction engine exploit these models for better predictions.  

For example, an officer may know that in one neighbourhood, break-ins follow a loose pattern, which they can describe in a behavioural model.  Instead of an experienced officer sharing robbery stories with a novice over coffee and hoping for an experience transfer, they can document their experience in behaviour models.  These models then feed an engine that can predict where and when burglaries may happen according to them and other sources of information.  More complex models may generate predictions based simply on the information coming into the system without a pre-cursor criminal act.

The Future

ISS’s DfuzePredict system takes the concepts in the 3rd generation of predictive policing tools and extends them even further by incorporating pragmatic engineering practices and years of lessons learned designing and maintaining intelligence systems.  Multiple engines are used in lieu of a single prediction engine or tunable algorithm because no single algorithm is expressive enough to predict different types of criminal
activities.  Also, crimes in one municipality may not follow the same patterns as in others.  In other words the activity patterns and sources of information available in a large metropolitan area may not directly translate into a those of a smaller city.  Breaking and entering is less likely to follow the same patterns as crimes of opportunity, therefore different algorithms, models or engines should be used for each type of criminal activity.  Multiple law enforcement organisations are being consulted to build a library of these models and engines for a wide array of crimes in varying municipalities as part of the implementation of these next generation platforms.  It’s definitely a situation where one size doesn’t fit all.

Bookmark and Share
 
Home I Editor's Blog I News by Zone I News by Date I News by Category I Special Reports I Directory I Events I Advertise I Submit Your News I About Us I Guides
 
   © 2012 DefenseFile.com
Netgains Logo