A high-tech team of number crunchers received a $10,000 prize in an Arlington competition aimed at examining public school student data to predict dropout rates.
The Big Data Roundtable event drew 23 teams, including the winning group from Deep Learning Analytics, which had experts from academia, the World Bank and top consulting corporations such as AIG and Booz Allen Hamilton.
John Kaufhold, a data scientist with Deep Learning Analytics, said the team used a massive cache of student data, including test scores, grades, attendance, courses taken and demographics. The trove helped the Deep Learning Analytics team find “how early warning signs in this data may identify students” at risk of dropping out, Kaufhold said.
“We hope this challenge marks the beginning of a continuing process to better leverage state-mandated data to improve student outcomes,” Kaufhold said.
Superintendent Patrick K. Murphy said the competition was developed to find ways to improve the school system’s graduation rate, determine how to increase the number of students who pursue advanced diplomas and better prepare the county’s students for college or careers.
Murphy said teams were given access to 12 years of school data, with individual student identities kept anonymous. Analysts found that attendance history, class performance and socioeconomic status were the most accurate predictors of future dropout risks.
“It reinforces the research that’s already out there,” Murphy said. “At the end of the day, we want to see kids take paths that lead to successful futures.”
The competition began in November and stemmed from a conversation between Murphy and Raj Adusumilli, assistant superintendent for information services. Adusumilli said that he and Murphy were in the superintendent’s office discussing the recent presidential election and how big data helped campaigns assess voters.
“How can we do it?” Adusumilli recalled Murphy asking.
The result was the big data challenge sponsored by CK-12, a Silicon Valley-based nonprofit that seeks to improve access to education around the globe. CK-12 provided the entire $10,000 prize, school officials said.
Last year, the high-performing school system saw a 6 percent dropout rate for the class of 2013. Using data analysis can help administrators keep an eye on students who might need encouragement to stay on track to graduate, Murphy said.
The data also can help the school system identify trends, including that students who struggle in certain math courses tend to also struggle in foreign language courses later on, Murphy said.
Aneesh Chopra, an Arlington schools parent who served as President Obama’s first chief technology officer, was a volunteer adviser for the project. Chopra said the central question of the event was, “Are we doing everything we can to extract insights from that data?”
He said that unlocking “the value and insights that come from it” ultimately will allow the school system to “help kids be prepared for success.”