At Big Mountain Data, we believe that we can scrutinize the big data that surrounds the phenomenon of domestic violence and family abuse to find answers to solving this hidden-in-plain-sight national tragedy. With domestic violence affecting 75 women every hour, you have big data. It’s a problem at scale. One comment we hear consistently when we’re talking to people in the field – at every level and type of organization – is that the data is a mess, so it’s hard to tell what’s working and what’s not in the fight against domestic violence.
September’s media circus associated with the NFL Ray Rice scandal, highlighted many experts and programs around the country. In a 5-second clip on ABC’s This Week, I discovered Police Chief Marty Sumner of High Point, NC who said:
“In the five years before we began this [program], we had 17 domestic-related homicides. In the five years since, we’ve had only one.” – Marty Sumner, Chief of Police, High Point N.C.
I listened to the clip again and again. WHAT the HELL were they doing to put up results like that? I had to find out.
So, I wrote to them.
I received a wonderful response from the department that included two downloadable PDFs that explained High Point’s offender-focused deterrence program. This approach was exactly the approach we wanted to focus on – offender-based strategies. I devoured the PDFs and did further research. Soon enough, I had some ideas of my own how we could even improve upon what they were doing, and achieve greater exposure for their success story.
So I reached out to them again. “Can we have a conversation?”
They liked our ideas and agreed to engage with us collaboratively in an online social network to gather ideas around projects and various initiatives. I brought in outside experts, and connected with their partners. The first project we’re engaging on together is coming up this weekend in San Francisco.
I’m (more than) pleased to announce The High Point Police Department is now included in the inaugural Hackathon for Bayes Impact, a prestigious Y Combinator-backed nonprofit that applies Data Science for Social Good. Our project is competing with The Gates Foundation and The White House. How cool is that?
We are supplying four, rich datasets for the data science teams. We are looking for one specific insight and one more general one. As it turns out, the officers on the ground have a hunch on some key indicators that may lead to repeat domestic violence. I choked up when Captain Tim Ellenberger said, “If we can intervene and deter the offender at the precise moment before the first arrest is made, we can prevent the cycle of violence from ever beginning.” The data scientists will be able to see this indicator in the data. Our second prompt is focused on the subgroup of repeat offenders and explores internal and external datasets to see what correlations exist that may identify actionable markers.
One of my favorite flicks of all times is, “The Butterfly Effect.” I know it’s not an award-winning film, but the notion that one could go back and correct a devastating moment in history is a fantasy every victim entertains. Behavior does not happen in a vacuum. There are triggers, forces, and sets of circumstances that can be analyzed as discrete data sources. What used to be considered science fiction is now possible by identifying behavioral patterns that can prevent a lifetime of harm, and can even save lives.
It’s very exciting. If Big Mountain Data closed today and this was ALL we did, I would celebrate heartily. But, of course this is not all we’re doing. This is day one in our #fightback strategy of preventing family violence from stealing the lives and sanity of innocent victims. We are fighting domestic violence with math and science.
Special thanks to Ian Thorpe of the United Nations who offered the very cool prize of a private tour of the United Nations in NYC to the winning data science team.
2 thoughts on “Big Mountain Data Heads to San Francisco!”
Policing is reactive, rather than proactive, the analysis of the demographics is going to be done how? How will you pro-actively identify cases of domestic abuse before they happen ? In Women violence and social change by Dobash and Dobash they show that only 2% of domestic violence goes reported, thus 98% of crimes have no data recorded against them and therefore the analysis of the data for any algorithm will be skewed towards the reported data rather than the unreported data.
Again, prior to any crime being reported, the domestic situation where there is a potential for a crime ,will not be in the public domain and thus not available for analysis to create or extend theories which in turn will be used to determine observable markers.
There is also the danger that individuals who have not yet committed a crime and thus still innocent may be flagged and targeted as such. How will you be dealing with that?
Hi Jamie. Happy to see you understand the magnitude of the problem. You are correct about the large number of cases that go unreported. We’re not able to study data we don’t have. We’re focused on the “known knowns.” In this specific case related to the hackathon data, the offenders are known. We are poring over the data to cull clues about behavior and correlations with other datasets. There is no danger to individuals who have not committed a crime. We’re not extrapolating from this dataset to the larger population. We are interested in reducing recidivism among the offenders already in the dataset. Thanks for your comment, and we’ll let you know how things go. Also, the focused deterrence strategy takes a proactive approach to reducing crime and violence. It’s a novel and effective approach that has yielded ground-breaking results.