PublicEngines Logo

Find Us On

Follow Us on Facebook Follow Us on LinkedIn Follow Us on Twitter Follow Us on YouTube Follow Us on our RSS Feed

Connect with Your Citizens Anywhere They Want - CityConnect: New Mobile App for Law Enforcement

CommandCentral Predictive: Built for Field Use


We have had tremendous feedback since we announced our newest product, CommandCentral Predictive, several weeks back.  Since the announcement we have literally heard from hundreds of agencies.  Last week I was on the road speaking to agencies, including several of our existing customers and hopefully some new ones.

It is hugely rewarding to see such response given the amount of time and effort we have put into creating CommandCentral Predictive.  This launch is the culmination of months of meeting with customers and getting their continuous feedback on the product, and the extraordinary effort by our team to turn that feedback into tangible changes in the product.

During the course of developing this product we spoke with many law enforcement agencies about product features and had dozens of recommendations on how to build a better product.   Some we have been able to include in the first generation, and others that are still to come.

From my meetings in the last two weeks there were two things that have really stood out about CommandCentral Predictive feedback.  First is that every agency that I have spoken to is looking for a better way to direct and enable their patrol officers to increase their effectiveness.  It’s simply an issue of how to maximize their resources in the most effective way through direction and information enablement.

The second item goes along with that was something that we have heard all along: that anything we provide needed to be usable in the field, not just another screen on a computer for someone to look at.

That hit home last week when I visited a large agency who is currently testing their own internally-developed predictive analytics solution.  While they have yet to determine whether or not to roll it out, one of the concerns they had was whether or not officers will actually use it.  Their limited trials have had mixed results and previous experience with another application showed limited success because very few officers got to the first step: logging on.

We realized this early in the development of this product which led us to two needs: first, to build CommandCentral Predictive with mobile devices in mind, and second, to empower officers with real data to make smarter, more informed decisions.

The mobile part was a no-brainer.  If they haven’t started already, almost every agency envisions the time when they will drop their MDT’s (Mobile Data Terminal) for tablet devices.  So, we built this product not only to be run on a tablet or large smart phone, but to be optimized for it.

Second, we built functionality into it so that users could access the underlying data driving predictions.  And it was designed to present this information to officers so they could easily access it and use it to make better decisions when they are in the field.

This data accessibility was validated time and again last week as I spoke with agencies, who see the value in empowering their officers with data in the field that help them to make better decisions and actions.  It simply helps to make better, smarter decisions while on patrol.  And, most importantly, it does it automatically, improving information flow without burdening an analyst with pulling data for officers (more on this another time).

Anyone who takes CommandCentral Predictive for a test drive will see how easy it is to access information that will put an officer at the advantage.  It’s just one of those things we built into CommandCentral Predictive that makes is such an exciting product.  We look forward to more feedback as we meet with customers and continuing to innovate on CommandCentral Predictive to make it the most effective tool to predict and prevent crime.

The Science of Predictive Policing.

Author: Praneeth Vepakomma, Data Scientist, Public Engines

The scientific problem of being able to make predictions has raised the interests of researchers for centuries. Solutions that attack this problem have found all encompassing applications within various fields of societal, applied and theoretical interests. Over this period, the complexity of these predictive models gradually increased and so did the predictive accuracy and generalizability of these solutions improve with time. Stemming from the early work of Gauss in the early 1800’s to Andrew Ng’s present day experiment at Stanford consisting of autonomous helicopters that learn to fly- the problem of prediction has truly moved from the simple problem of fitting a straight line, to learning complex relationships from data.

R&D at PublicEngines

At PublicEngines we have a committed R&D team focused on this intersection of developing advanced mathematical models internally for predicting crime that also ‘learn’ from data and we are pleased to have announced our newest product, CommandCentral Predictive that culminates as a result of these dedicated research efforts.

Mathematical Modeling of Crime

The mathematical model that I have developed in-house along with my team exploits specific behaviors that are markedly inherent to crime-incident data. These behaviors include near-repeat victimization, long-term patterns, transient/short-lived patterns and interactions within crime-incident data that our model understands while separating signal from the noise thereby raising the statistical confidence levels around our predictions. These dynamic sets of crime-specific behaviors are measured through our model while not just separating one characteristic from another but also while accounting for the dependencies between them to mathematically understand the inherent predictive characteristics of the patterns of crime. We also were cognizant of some inconsistencies that crop up through the use of some of the existing academic literature around similar problems, and improve upon these as well within our patent pending system.

Focus on Models that Learn

The automated learning component of our model is another important aspect that brings in generalizability to our predictive engine where it guarantees a level of robustness, over new unseen data, when the models are deployed in reality. The scientific field of building various domain specific models that also have the ability to learn from data is an integral part of this field called statistical machine learning. This area began to gain early momentum, when scientists began to build mathematical models that mimic neural networks within the human brain, starting from late 1940’s. An early learning-system breakthrough occurred when scientists in the late 1980’s were able to develop handwriting recognition systems. The field grew leaps and bounds from there on, thereby adding analytical intelligence to various use cases. Today’s E-mail Spam Classification Systems, Voice Recognition Systems, Computer Vision Systems, Facial Recognition Systems, Document Classification Systems are just a few examples of the successes of machine learning. The main reason to focus on a learning component is because in reality, a model that works great over the data that you use to build and train the model on, wouldn’t always be replicable in new instances of these datasets when deployed in real-life scenarios. This happens primarily because the model could learn non-generalizable intricacies within the training dataset, which do not otherwise translate into practical results over future data, that the model would face after deployment. This is the reason; we rigorously train our patent pending crime-specific statistical model, and as well have a learning component comprising of an ensemble of multiple models to keep it robust in real-life deployment. This allows for our model to separate the predictive characteristics of future crimes, down to a high level of tactical/actionable granularity.


Field-testing & Evaluation Metrics

We have rigorously tested our models against the most traditional methodologies like ‘Hotspotting’ which are heat-maps generated using historic crime-incident data. In order to have a scientific evaluation, we quantify the performance of existing Hotspotting methodologies by overlaying a grid over the hotspots and measuring the success rate of our predictions in comparison to the hotspot. The number of days required for an evaluation ranges somewhere between a minimum of 80-120 days based on the dataset in order to reach a statistically significant comparison. We conclusively observe, that our patent pending model performs about 2.7 times better than the widely used hotspot.

Positive Impact of Actionable Patrol

This level of focused tactical information would help reinforce ‘directed, actionable patrol plans’ and increase ‘resource-efficiency’ so that agencies can use it in their law enforcement efforts to positively impact their communities through predictive policing. This technology unlocks tactical intelligence from your RMS and closes the analyst/officer gap by efficiently improving operational decision-making. From the perspective of the tech-industry as well, this is absolutely inline with our goal of continuing to reinforce officers with technology that provides tactical and directly actionable information in order to assist them in continuing to do their best.


Predictive Analytics Where it Matters: Preventing Crime

Screen Shot 2013-08-06 at 10.58.55 AM

Today there is a lot of buzz about the use of predictive analytics in business.  Spurred in part by the best selling book, Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie or Die, by Eric Siegel, it seems that everyone is talking about ways to predict every major and minor event in our lives.

As with every application of technology, there are implementations that can have a higher, or lower impact on society.  At PublicEngines, we have long believed in the application of analytics to improving quality of life in general, and specifically to fighting crime.  That’s why we were pleased to announce our newest product last week: CommandCentral Predictive.

CommandCentral Predictive is a significant step forward in the use of predictive analytics to accomplishing law enforcement’s goal of preventing crime and improving our communities by making them safer.  In particular, from the time we conceived this product through to its final development, there were two things that were most important to us: accuracy and ease-of-use.

Pre-eminent in our thought process was to create a product that accurately predicts potential crime.  The entire basis of this product is to more accurately digest data and provide an accurate, unbiased view of the highest probability of crime on a daily basis.  So, the science behind it had to be sound.   That’s why we designed our prediction engine using multiple algorithms to optimize its effectiveness and verified it with extensive field-testing.

Of almost equal importance to us was to design the product with the highest ease of use.  It doesn’t matter how good the product is, if the interface isn’t easy to navigate,  it won’t be used.  So, we designed the product from the start with the idea that users need to be able to jump in the product right from the beginning with little to no training.  Then, we took that design to officers, analysts, and command staff, and asked for their input on how to make it better and we redesigned it based on their feedback (more on this in another post).  The result is something that is functional and highly usable; delivering a daily report in a way that any officer can use to improve the way they police.

With an announcement of a product like this there is no doubt that naysayers will voice their opinions.  In particular, we’ve heard those who will say that advanced crime analytics software can’t replace crime analysts.  And they are right.  That’s not our intent.  However, we know how much time analysts have and where their demands are.  Analysts spend a significant amount of time working with officers and patrol supervisors on their areas of patrol responsibility.  They’ve told us they’re overwhelmed with more work than they can handle.  CommandCentral Predictive is designed to help them take a significant demand and essentially automate the prediction of high-probability events and give them more time for analysis.  For those agencies without an analyst (a majority), this is a significant boost by giving them the tactical, directed analysis their officers need but can’t currently afford.

Others may cite instances where software has not worked well at making predictions, such as the military’s use of prediction software to forecast political unrest.  But what we have seen is that, like what you purchase, where you shop, or what you look at online, crime occurs, for the most part, in a repeatable, predictable pattern.  And our field-testing has shown that our algorithms are far more accurate in seeing both long- and short-term trends, modeling them, and learning and improving along the way.  In fact, this very field-testing has show us to be, on average, 2.7 times more accurate than traditional hotspot models at determining where the next crimes will occur.

Tim O’Reilly, the well-known media technologist, once said, “We’re entering a new world in which data may be more important than software.”  I think this is especially true in law enforcement, where enforcement officers have the data to help themselves, but have traditionally struggled through poor software tools to help them analyze it.  So, while we are proud and excited about CommandCentral Predictive, what we are most excited about is this product’s potential to unlock the patterns and intelligence in agency’s data and helping them to make better and more effective policing decisions.