Data Innovators Darrin Lipscomb

Published on February 23rd, 2016 | by Joshua New

0

5 Q’s for Darrin Lipscomb, Senior Director of Public Safety for Hitachi Data Systems

The Center for Data Innovation spoke with Darren Lipscomb, senior director of public safety for Hitachi Data Systems. Lipscomb discussed how predictive policing technology can improve how police departments fight crime, as well as how he thinks crime modeling will eventually be as common as weather forecasts.

This interview has been lightly edited.

Joshua New: At Hitachi, you work on developing crime monitoring and prediction technology. How exactly can software predict crime?

Darrin Lipscomb: There’s a huge amount of public data available to leverage. The big challenge for law enforcement is turning all that data into meaningful insights—correlating disparate data sets and making sense of them to not only gain insight, but ensure those insights are actionable.

At the highest level, there are three necessary components: first, the ability to rapidly ingest data from variety of sources and ensure that data stays current;  second, a flexible analytic model that enables input of features from the overall data set, such as a historical crime series, so it can then perform statistical analysis to assign a granular threat level, as well as weigh underlying risk factors; lastly, it requires a way to simply and intuitively visualize the information.

The first component provides the ability to ingest various disparate data sets and also has the ability to continually update this data in order to ensure its relevancy. This is a key piece of the puzzle and often the most difficult for public safety organizations to achieve in a near real-time manner. Our software, Hitachi Visualization Suite (HVS), leverages Hitachi technology to ingest critical data from a variety of sources on a continuous basis and securely transmit it to a centralized repository in the cloud.  

The second component is the model. At Hitachi, we opted to leverage a multi-variant, spatial-temporal model, which essentially enables us to input any variable feature into the HVS system where we believe there is a cause and effect. For example, if our crime series is of sexual assaults and we believe there may be a correlation to known sex offenders, we can selectively add sex offender registries to our analytics model as a feature. It will then assign a weight to the feature so the user can quickly determine if there is a correlation. In addition, it will assign a threat level to every city block—we call this the threat surface.

Lastly, HVS provides a highly flexible geospatial view for visualization of the threat surface.  Darker colors represent higher relative threat levels and the user can choose a threshold to, say, only show the top 25 areas of the city with the highest threats of sexual assault for example. We can also show other assets such cameras and historical events which have occurred in and around these areas. So overall, we are providing the tools that allow public safety organizations to become much more proactive.

New: It’s not hard to see why certain data sets, such as history of crime in a particular neighborhood, would be useful to this kind of technology. What kind of data can help predict crimes that might not be so immediately obvious?

Lipscomb: Law enforcement may have assigned gangs to specific anchor neighborhoods. This data can be very useful to determine threat levels based on gang locations. In addition, other data such as weather forecasts, and economic or demographic data may also have a correlation to specific types of crime and can certainly be leveraged in the model.

New: Continuously monitoring and predicting crime must be a pretty big analytical challenge with all the data involved. How do you make sense of all this data?

Lipscomb: More data simply requires more compute cycles and we leverage virtual machines in the cloud to achieve this. But the job of the model is to work alongside a human being so it can be continuously modified and improved upon to become an increasingly useful tool for that person over time. This is why understanding the underlying risk factors are so important. When we get to dozens of variables, the software needs to have the ability to tell the user if a particular variable has an effect on the probability of a crime occurring.

New: Every time predictive policing is mentioned in the news, comparisons to the movie Minority Report are practically unavoidable and it paints a pretty scare picture of this technology, suggesting that it could lead to a dystopic police state or circumvent legal protections like due process. Has this association limited the ability for it to be put to good use?

Lipscomb: Our model really doesn’t target individuals, although one individual could have a huge impact on a particular type of crime, such as if a criminal was released from prison then all of sudden there are a rash of car thefts. This is information the model would need to have available in order to identify the correlation. Ultimately though, we are simply telling law enforcement where in the city are the highest probabilities of specific crimes based on real data and historical trends, so they can allocate their resources in a more efficient manner.

New: That’s not to suggest that this technology—like any technology—doesn’t have risks. What steps can law enforcement and city officials take to use this kind of tool effectively?

Lipscomb: It starts by embracing it. Public safety budgets are not keeping up with population increases and the overall threats we are facing. These types of technologies are going to be an absolute necessity now and especially in the future for both violent and non-violent crimes, as well as terrorism. As with technological innovations of the past, we need to follow the guidelines and policies that have been developed to mitigate the risks of their use and there are many organizations working on developing these guidelines and policies today. I firmly believe that in 20 years, crime models will become as commonplace as weather forecasts.

Tags: , , , , ,


About the Author

Joshua New is a policy analyst at the Center for Data Innovation. He has a background in government affairs, policy, and communication. Prior to joining the Center for Data Innovation, Joshua graduated from American University with degrees in C.L.E.G. (Communication, Legal Institutions, Economics, and Government) and Public Communication. His research focuses on methods of promoting innovative and emerging technologies as a means of improving the economy and quality of life. Follow Joshua on Twitter @Josh_A_New.



Back to Top ↑

Show Buttons
Hide Buttons