Eleven LANL data scientists worked full-time for a week on a joint project with Northern New Mexico College (NNMC) and Santa Fe Community College (SFCC), and with nonprofit Rocky Mountain Youth Corps (RMYC).
The colleges' project focused on improving student retention rates by analyzing risk factors and signs of academic and professional success.
“When you have an institution like us – with open admission – situated in a region that has generational poverty and socio-economic issues, students drop at a higher rate than an institution in a city that has high standards of admission,” says Ivan Lopez, provost and vice president for academic affairs at NMMC. “If we can know, based on their admission documents, what are the chances that a student is going to drop, then we can have early interventions and hopefully help that student succeed.”
Rocky Mountain Youth Corps provides workforce development training and educational programs to young people in New Mexico, and it was looking to understand its general impact, and its performance in gender diversity.
“Over the last few years we’ve been conducting surveys of the corps members, and we had all sorts of data in a range of spreadsheets,” says Herbert Foster, development director at RMYC. “But we did not have the capacity to analyze the data like the LANL team did, to draw some conclusions about what was working or not, and what were some of the determinants of success.”
Cleaning the data
“The first two days were focused mainly on data wrangling, which is a necessary but frequently thankless job, where you’re not really doing the analysis, you’re just cleaning up the data,” says Lissa Moore from the Laboratory’s information sciences group, who organized the sprint. “I think it was at our end of day check-in on Wednesday when everyone came to the group session excited that they were starting to get results, and starting to feel that sense of success.”
The LANL teams used several different techniques to analyze the data, and their final reports contained some findings that confirmed the partners’ prior impressions, and others that were more surprising.
For RMYC, the analysis suggested that while there was a gender disparity in the numbers of members served (as the nonprofit suspected), the women that did take part had positive and impactful experiences. The project also showed that returning corps members benefited less from the program than first-time attendees. Both results will inform where the organization focuses its recruiting efforts in future.
For the colleges, the analysis bore out at a local level the national findings that while other issues also play a role, financial factors were the most important determinants of whether students would drop out or not. Confirming this will help the institutions to direct financial aid resources to the right students early in their college careers.
More classes, more success
One more surprising finding was that students taking more courses were more successful than those taking fewer courses.
“For students taking fewer courses their chances of success might be lower in part because their external non-academic barriers might be greater,” explains Yash Morimoto, associate vice president for planning and institutional effectiveness at SFCC. “But for people on the borderline it might be helpful for them to take one additional class so they graduate as quickly as possible. That’s influencing our decision on how we can help people who are already taking a full-time credit load to add more classes for the same price.”
Setting the partners up for the future
The LANL team also ran workshops explaining how to do some of their analyses, and outlining readily available tools that would allow the partners to replicate some of the work in the future. They also gave recommendations on what sorts of additional data it would be worthwhile to collect.
The Laboratory researchers’ time for the week was paid for by the Laboratory’s Information Science & Technology Institute (ISTI) and the program is coordinated under the Lab’s Community Technical Assistance (CTA) initiative, which makes the unique expertise and capabilities of the Laboratory available at no cost to tribal entities, governmental entities, and nonprofits located in the seven counties of Northern New Mexico. The U.S. Department of Energy's National Nuclear Security Administration (NNSA) funds both ISTI and the CTA initiative.
As well as the practical findings and recommendations, the partners will take away some positive memories of the week.
“The sense projected by the LANL team was that they were excited to give back to the community, and that was a pleasure to see,” says Lopez.
“You always love to watch people who are experts in their field doing their work, no matter what area,” says Morimoto. “But it’s especially true for me as a data geek to watch these data scientists doing things that I’d only seen in the movies. It was just amazing.”