Making students create public tutorial videos as a course assignment
Many university courses require students to spend a lot of time writing reports and generating material that is afterwards never used for anything and remains in the private Dropbox account of the student. When you do the math, it is quite an impressive amount of knowledge which is created, solely for the sake of assessment. A different problem is that students often write their reports in a language addressed for academic people (the teacher) instead of other students.
When teaching Computational Tools for Big Data last year, we tried an experiment to address both problems at once. As the fourth and final assignment in the course, we asked the students to produce a video tutorial on a technical subject relevant to the course, and release this video on YouTube.
The requirements for the video were:
- You have to hand in a link to a Youtube Video
- You can create the video alone or in groups of 2 or 3 students
- The video should be public or unlisted if you are not comfortable with showing it to the world
- The video should be at least 5 minutes long, and at most 15 minutes long
- The video can be structured in any way you want. It can be an interactive tour in an iPython notebook, a slideshow with your voice overlay, a cartoon or whatever you can come up with
- The video should cover a subject relevant to the course, but not a subject that we already covered in the course. Think of the subject as “What would I teach if I had one more week?”
After the students handed in their videos they had to peer evaluate them through Peergrade. Each student was asked to watch and evaluate 5 videos. The videos were evaluated using the following criteria:
- Was the presented subject relevant in a Big Data context?
- Was the presented subject relevant in a Computational Tools context?
- How well explained was the presented subject?
- How difficult is the presented subject (how technically challenging is the subject)?
- How well-produced was the video (picture, sound, effects)?
- How good was the video overall (combination of subject, presentation, video production)?
Each subject had to be pre-approved by me (the teacher) to ensure that the subject was relevant enough and that we did not get too many videos with the same subject. This was handled through a Google Form.
What came out of it?
In total, the students of the course produced 45 videos that were released publicly to YouTube. The topics included for example: Google TensorFlow, Apache Kafka, Neo4j Mazerunner, Apache Solr and Docker.
The five videos which obtained the best evaluations from the other students were (not in order):
Interestingly the video which has since the course gained the most views was not in the top 5, but has so far been seen almost 6000 times! It is a video on how to train Convolutional Neural Networks with TensorFlow:
What is even more fantastic is that the students that produced the video have since the release helped YouTube-commenters by answering various questions.
You can find a YouTube playlist of all the videos created by the students here:
When we run the course this fall we will definitely try out this experiment again and probably change a few things to make it even better.