This post was written by Alis Wang
Gun violence has been a vexing problem for law enforcement and policy makers for decades. This is partially because of the difficulty of predicting which individuals will perpetrate the next mass shooting. Big data, however, could give us a way to objectively analyze this issue and find ways to reduce the number of mass shootings. Currently, the US lacks a national recordkeeping database of gun owners or firearm and ammunition purchases. Data science, however, could be used to build one to track and analyze gun and ammunition sales. This could provide law enforcement agents with better data and tools to flag potential violent perpetrators. For example, predictive models could be built with the gathered data to analyze trends in gun purchasers’ activities to distinguish non-offenders from those with a higher probability of committing gun violence in the future. Because of the historical data available on past perpetrators’ purchases, data analysts could build useful predictive models. This would improve our ability to flag suspicious behavior before potential killers commit an atrocity.
For example, James Holmes, the suspect in the Aurora movie theater shooting, displayed suspicious behavior prior to his violent attack. He went to three geographically spread out stores to purchase four different guns. Holmes then ordered assault gear and thousands of rounds of ammunition online; UPS subsequently delivered about 90 packages to his address within a short period of time. If data like this—in addition to background information on Holmes’ mental health issues—were available to law enforcement officials and data analysts, Holmes’ rapid buildup of arms and ammo would likely have popped onto their radar, potentially leading to action before he committed the tragic mass shooting.
In addition, developing a database of guns, ammunition, and their owners is actually quite trivial from a data science point of view and some authors say it does not even fall into the “big data” category. In fact, all of the information needed to build a national database could fit on a single thumb drive and could potentially even be analyzed using traditional data mining and business intelligence technology. If data scientists are provided with clean and consistent data, building these predictive models and analytical databases would not be difficult at all.
What is preventing progress on this database is therefore not the technological skills required, but rather concern over citizen’s privacy rights along with a lack of political will and interagency cooperation. Political will and cooperation between different authorities is necessary in order for analysts to obtain the permission to gain access to this data and to build a national database. Privacy concerns are perhaps the most important limitation, especially as data becomes more readily available. This month, for example, Journal News unleashed a maelstrom of criticism when it published the locations of 44,000 registered gun owners in New York State using Google Maps even though the newspaper obtained the information legally under the Federal Freedom of Information Act. While a national database would be able to help anticipate gun violence, this must be balanced against an American citizen’s reasonable right to privacy. This is sure to be an ongoing political debate, and data analytics should be part of it.