Abstract:

This paper will present an integrated learning, practicing and assessment environment for courses in computer science, math, statistics and other fields that is based on a central installation of JupyterHub. By extending the functionality and using the full potential of JupyterHub’s configuration options, it supports the entire lifecycle of the teaching and learning process. This includes the production of interactive lecture slides and lecture notes from a single source, but also the design of problem sets (assignments) that can be automatically rolled out to all students of a course or to specific groups of students within a course. Once these problem sets are due, student submissions are automatically graded and feedback is sent back to students.

This allows to conduct a problem based learning approach also for huge groups of students, since the effort for grading submissions is minimal. Some shortcomings of JupyterHub – like missing or poor support for multiple courses, multiple teachers or multiple groups per course – have been fixed by custom extensions and scripting. Also the technical infrastructure, which is formed by an on-premise Kubernetes cluster, is quite sophisticated. It is highly scalable and resilient and also supports customized resource allocation based on the requirements of different courses.

Keywords:

Jupyter, JupyterHub, Kubernetes, e-learning.