Project-Based Learning in a Machine Learning Course with Differentiated Industrial Projects for Various Computer Science Master Programs

Raphael Brüngel, Johannes Rückert, Christoph M. Friedrich

Abstract: Graduating computer science students with skills sufficient for industrial needs is a priority in higher education teaching. Project-based approaches are promising to develop practical and social skills, needed to address real-world problems in teams. However, rapid technological transition makes an initial training of contemporary methods challenging. This affects the currently much-discussed machine learning domain as well. The study at hand describes a re-framed teaching approach for a machine learning course, offered to various computer science master programs. Project-based learning is introduced with differentiated projects provided by industrial partners that address the diverse study programs. Course attendees are supported with manuals, tools, and tutoring, passing through the Cross Industry Standard Process for Data Mining (CRISP-DM). Observations made during two iterations are reported, accompanied by a first empiric evaluation of student experiences.