Week 2: Applying AI to the Student Debt Crisis

by Nancy Rubin, ASU ShapingEDU Innovator in Residence

Week 2 of the “Applying AI to the Student Debt Crisis” project was an exciting week. The teams began diving into work, defining tasks and identifying working groups.

After the first project team call last Sunday, questions emerged from participants. Many of the Omdena data scientists working with the ShapingEDU educators are from other countries and are not as familiar with the nuances of student debt in the United States. Based on the questions posed, I created a presentation and we scheduled a second weekly call for anyone available to answer some questions.

Slide with a couple bullet points about student debt in the United States

I shared additional information about student debt, identified current data sources that are publicly available and asked everyone to keep track of the sources of their data and information from so we can create a bibliography later.

Sunday, on the weekly project call, over 40 participants attended to receive updates and provide their input on their progress. Participants ask questions, get feedback and are mentored through the process by the Omdena team.

As we dive deeper into the problem of student debt, some specific areas of focus are emerging.

  • Examining demographic data and earnings to understand loan repayment patterns.
  • Why is student debt unevenly distributed? Minorities and first-generation students are often more impacted by loans.
  • Exploring spending habits of students and parents and using data from the Consumer Expenditure Survey.
  • Mining social media sites for sentiment analysis on the debt crisis, loan forgiveness, and generalized information.
  • What is the impact of debt on mental health?
  • Is there a group of students or parents being targeted by predatory loan providers? Can we segment loan data to better understand the different loans and who they are accessible to?
  • Is there data available on university websites that could contribute to our understanding of the problem? Can we scrape the data from the websites?
  • How is the problem in the United States compare other countries? Are the data sets available from other countries we can use to do comparisons?
  • In a post-Corona world, will there be more loan defaults? What are the economic impacts in different sectors? Can we create predictive models so students can make informed decisions about degree programs?

It is exciting to be a part of such a fast-moving project. Next week, groups will begin presenting their initial findings so the rest of the group can provide feedback and find synergy among the group and tasks being worked on.

More next week!