Personalized Transfer Guidance for Community College Students

A machine-learning algorithm to provide transfer-intending community college students with a personalized list of courses that meet degree requirements and maximize their predicted probability of academic success.

The Problem

  • 80% of community college students plan to transfer, while only 20% actually do so.
  • Students have very limited access to advising with student-adviser ratios often greater than 1,000:1.
  • Existing articulation agreements between community colleges and 4-year institutions are dense, complex, and hard to decipher.

80% of community college students plan to transfer, while only 20% actually do so.

Deep Dive into the Data

A full overview of the algorithm, with details on modeling choices and variables we include, is available here. When we compare the predictions from the algorithm against the status quo we observe in terms of student course outcomes, on average we predict students would :

  • Earn 2.73 more credits that count towards VCCS degree reqs each semester.
  • Earn 2.90 more credits that can directly transfer to  @uva per semester.
  • Increase GPA by 0.19 points per semester.
Click to enlarge.

The Innovation

  • We developed the algorithm within a research-policy partnership with the Virginia Community College System 
  • This is a beta version of the algorithm and we’d love feedback on how we can strengthen it.
  • We are working with higher education institutions and system leaders to gather feedback with plans to pilot the algorithm in the context of intrusive advising intervention in 2020.

The Project Team

Ben Castleman, Founder and Director, Nudge4
Catherine Finnegan,
Assistant Vice Chancellor for Institutional Effectiveness

Kelli Bird,
Research Director

Yifeng Song,
Chief Data Scientist
Zachary Mabel,
Associate Policy Research Scientist

Our Funders