Hitting the Sweet Spot: Designing an Accessible Introductory Machine Learning Course for Informatics Students

Machine learning applications are ubiquitous in today’s postmodern societies, with applications in healthcare, criminal justice, urban planning, envi-ronmental sustainability, etc. This rising interest in machine learning (ML) is matched by an ever-increasing demand among college students to learn ML. Unfortunately, the most basic and commonly used treatment of ML in computer science curriculums around the country is quite technical and math-heavy, which demotivates students from nontechnical backgrounds to take up ML courses (despite their interest in learning how to use ML). Students in iSchools (or information schools) are characterized by their diversity of educational backgrounds and a great commitment to interdisciplinarity. As such, an ML course offering inside an iSchool cannot assume prerequisite mathematical ability among all students who wish to take this course. This paper discusses an attempt at designing an easily accessible ML course that was specifically designed for Informatics students at the Pennsylvania State University. In particular, this paper discusses an important trade-off between a math-heavy treatment of the course and a more use-oriented treatment of the course. Further, this paper sheds light on the design choices made during the curriculum design, evaluation rubric design, and content delivery of the course in order to help balance this trade-off and, in turn, cater to the diverse backgrounds and interests of Informatics students.