Data Science Offers New Tools for Understanding Foster Care Outcomes

Daniel Oldham in front of the Student Union
Daniel Oldham, Embry-Riddle Class of 2019, leveraged data analytics techniques to shed light on the experiences of children in foster care. Photo: Katya Rivera

An analysis of records regarding Florida children in foster care between 2010 and 2017, completed at Embry-Riddle Aeronautical University, suggests a new way to inform decisions about child welfare services.

The data analytics project – conducted by student researcher Daniel Oldham, under the direction of his faculty mentor Dr. Mihhail Berezovski – revealed intriguing preliminary insights to children in foster care and their caregivers. 

For example, Oldham found that foster children with an older caregiver as well as children who are younger seemed to have better success in exiting the foster care system. Also, those with longer-duration cases – indicating they did not “bounce” in and out of the system – seemed to have better outcomes, Oldham reported. Moving from foster care into a relative’s home was also associated with success, he added. 

Someday, “predictive analytics” methods could set the stage for policies to help more children successfully exit the foster care system, said Oldham, who will earn his Embry-Riddle bachelor’s degree in Computational Mathematics on May 4. 

Oldham’s research – requested by Partnership for Strong Families, a child welfare services organization in Gainesville, Fla. – focused on foster-care data from the Florida Department for Children and Families (DCF). The DCF records, encompassing 250,000 records from 40,000 cases, included information such as children’s ages, locations, case histories, services provided and clinical records. 

As a first step, Oldham wrote a mathematical program that automatically scanned for key features in the massive dataset in order to calculate the “weight” of each case. Lower weights were associated with better outcomes for children. 

“I had to create a scale to show successful versus unsuccessful foster care outcomes,” Oldham explained. “Each potentially negative event in a child’s life, such as relocation from a foster home to an institution, added weight to the case. Each positive event, such as being adopted, reduced the weight, indicating greater success in exiting foster care.” 

Oldham then leveraged a neural network and machine learning to determine factors that influenced foster care cases in a positive way. 

New Master’s Degree in Data Science 

This fall, Embry-Riddle will launch a Master’s degree program in Data Science, to encourage students like Oldham who want to develop solutions to real-world problems. 

“As soon as my students graduate and enter the job market, they will be in a world full of data,” said Berezovski, assistant professor of Mathematics. “I want them to have the skills to conduct industrial mathematics research.” 

The new Master of Science in Data Science will be an interdisciplinary program that will prepare students to use the latest computational and analytic tools to solve data intensive problems for business, industry or government, Berezovski said. The goal of the program will be to provide the students with knowledge and skills in the areas of data collection, management, analysis, visualization and interpretation of large datasets associated with the various domain areas, he added. 

Regarding Oldham’s project, Berezovski said, “From a simple data collection, he extracted very useful insights that were otherwise hidden. He had an opportunity to work on a real problem and provide information to benefit society. That’s inspiring.” 

Oldham presented his results at the National Conference for Undergraduate Research in Atlanta. He also took part in Embry-Riddle’s recent Discovery Day, a showcase of outstanding student research projects, organized by the Office of Undergraduate Research. This research was sponsored by that office, through the IGNITE grant.

Posted In: Applied Science | Computers and Technology | Research