2025 ASEE Annual Conference & Exposition

The GradTrack Scholars program prepares undergraduate students for graduate school while building a community of students excited to pursue advanced study. GradTrack uses mentoring circles – a proven model for supporting individuals pursuing graduate school [1], postdoctoral roles [2, 3], and faculty careers [4]. In GradTrack, each mentoring circle unit consists of two graduate student mentors and 6–8 undergraduate mentees, where each unit is part of a scalable mentoring system. In 2024, GradTrack included 26 mentors and 100 mentees. Currently, the manual process of matching mentors and mentees is time-consuming, requiring 4–8 hours of administrator-assigned matching per program. While tools exist for grouping students into teams for class projects [5], most widely used tools are commercial, fee-based, and/or are more complex than needed. Therefore, we identified a need for a simple, open-source solution specifically for mentoring structures.

The purpose of this research project was to develop a streamlined method for the formation of mentoring circles as GradTrack continues to grow, specifically by leveraging widely used statistical algorithms such as k-means clustering. Using data from three years of manually created mentoring circles, we developed a python algorithm which uses groups of mentor pairs as seeds for clustering mentees. To do this, k-means clustering and one-hot encoding were used to create balanced groups of mentors and mentees based on similarities of interests and a specified number of groups. The output of this python algorithm is a list of mentees grouped by similar interests and mentor group number.

This study evaluates the benefits and limitations of using a computer-based and automated method to assign mentees to mentoring circles, and shares the process of development as well as a protocol for use. The key implication is a more efficient process for creating mentoring groups with a reduction in human time and subjectivity, which will support the continued growth of the GradTrack Scholars program and other mentoring circle program structures. To the author’s knowledge, this is the first study to develop a k-means clustering algorithm applied to mentoring purposes. Future study should evaluate the comparison between manual and algorithm based mentoring group formation in connection to assessment of mentoring group success.