On June 4, the Data Science Working Group (DSWG) came together with 80 civic hackers to attend SF’s National Day of Civic Hacking at the Microsoft Reactor. This year marks the 4th annual occurrence of this event, a nationwide initiative to build technology for social good. Promoted by Code for America, local brigades shared their experience through the hashtag #HackforChange.
In the morning, the DSWG partnered up with the California Department of Justice (DOJ). The DOJ’s Chief Data Scientist, Sundeep Pattem, along with his OpenJustice fellows, Rob Kuvinka and Sai Dulla, shared their insights around the types of policies the DOJ has jurisdiction over, as well as the statewide changes that were implemented throughout the years.
At noon, the teams broke out into three smaller groups to solve a range of problems:
- Data conversion scripts to enable more efficient integration of DOJ data
- Data-driven visual storytelling for the DOJ website to engage the public
- Predictive modeling focused on factors influencing arrests and dispositions
The visualization team achieved quick wins over the weekend. “I get to learn new technology,” said Han Seoul-Oh, a visualization volunteer. “For example, you can use Jekyll to make Github do fancy processing. Perfect for a hackathon.” Other members of the visualization team included volunteers Sanat Moningi and April Wensel, along with OpenJustice’s Kuvinka and Dulla.
As the day progressed, the DSWG and DOJ dived deeper into the modeling problem with 3 key prompts:
- Predict initial juvenile justice contact
- Predict juvenile recidivism
- Improve resource allocation for public safety
Coming out of the event, DSWG volunteers – Sushma Y., Peter James, Holly Davis, Jude Calvillo, and Catherine Zhang – continue with hypothesis testing and modeling of contextual and demographic factors that best predict juvenile felonies. They regroup with the DOJ weekly at SF Brigade Civic Hack nights. We invite you to follow our progress here: visit our Github repo >>