TU Wien:Visual Data Science VU (Schmidt)
|Alias||Visual Data Science (en)|
|Department||Visual Computing and Human-Centered Technology|
|Mattermost||Register • Mattermost-Infos•|
- Introduction to Visualisation
- Advanced Visualisation Techniques
- Introduction to Human-Computer-Interaction
- Introduction to Visual Data Science
- Usage of Charts and Plots in Data Science
- AI in Visualisation
- Trust in Visualisation
- Applications & Libraries
What is nice is that often the presented material is backed up by hints to literature (e.g. that the visualization presented, and the best practices for using it, are backed up by a scientific study about how to properly use it, etc.)
Weekly lectures until the end of December. Additionally, two practical exercises, one presentation and one TUWEL Test have to be done. There is no mandatory attendance for the lectures.
Weekly lectures with well prepared slides. Material is presented rather slow (not in a bad way) and thoroughly.
Different "grading packages" available: Either do:
- two practical exercises with datasets, visualize and analyze data and submit reports + presentation in the end.
- 1 practical exercise with dataset, as above. Then a final oral exam.
Attendance is checked with list, depending on grading package a certain number of attendances (5 to 7) can earn you points toward your final grade.
Knowledge of Python, especially Pandas, and visualization tools such as matplotlib, seaborn, etc., is necessary. For assignment 2, the usage of a tool or library for interactive data visualization is also required (or needs to be picked up).
Some statistical knowledge is also required, and basics of visual data analysis.
Lab 1: A dataset about food content is given in .csv format. The data has to be analysed statistically as well as visually. The topics were clustering, correlation and a comparison of three types food types. Additionally, a report of at least five A4 pages (with images) has to be handed in.
- Assignment ("Lab") 1: Given data-set, compare computational and visualization methods. Tasks: Cluster analysis. Check for correlations. Compare groups of variables. Written report of at least 3 pages is to be submitted.
- Assignment ("Lab") 2: Search your own large data-set. Explore, get insights. Then either: make a Dashboard for exploration, or: write report about analysis of 6 different visualizations, i.e. a literature review/survey.
The grading is based on the two lab submissions, the presentation and a TUWEL test. There are a total of 100 points: Lab 1 (20 points), Lab 2 (30 points), presentation (30 points), TUWEL Test (20 points). The TUWEL Test contains 10 random questions and can be taken multiple times. The best result of all tests counts.
Dauer der Zeugnisausstellung
- Lab 1: A few hours for the analysation of the data and the creation of the visualisations. Dependes largely on the knowledge of python and the needed libraries. A few more hours for writting the report.
- Lab 2:
- Test: Since the test can be taken multiple times, not much preparation time is needed. The questions are more practical (e.g. what kind of data can you see in the visualization? [Univariate, Bivariate, Trivariate, Multivariate])
- Presentation: The presentation is done with the dashboard alone (no slides). Maybe take a bit of time to think about what you want to say/show during the presentation.
- Assignment 1: a few hours, to familiarize yourself with the data, the tasks to perform, and the tools you need. Depending on your skills in tools for visualizations (some plots have to be created and analyzed) as well as methods for clustering (e.g. scikit-learn), your milage may vary. Then a little more effort for writing min. 3 page report (including figures where appropriate).
- Assignment 2: [can't really say, I switched to taking the exam rather than doing the assignment 2, as I coulnd't afford taking the time; it seems that taking the exam is much less effort than doing the assignment, but that is only my humble opinion. Please add other experiences]
- Exam preparation: 2-3 days (without attending the lecture: most of the stuff (diagrams) has been discussed in other courses (charts, perception, ...), for some other topics (AI Vis, charts/apps, ...) the referenced papers have to be read since the slides are bit sparse)
Slides are available from the classe's web page, see link above.
Links for data for Lab 2:
- Do not underestimate the time needed for Lab 2. You will probably have to spend quite some time on analysing the dataset itself, finding the insights and getting the visualisation and interaction to work.
Verbesserungsvorschläge / Kritik