After completing the course the student will:
- have an overview of the state-of-the art visualization methods for relational and high-dimensional data (Know);
- be able to explain how these methods construct visualizations (Understand)
- select an appropriate data visualization method and parametrize it when presented with high-dimensional or relational data (Apply)
- assess the results of the constructed visualizations (Evaluate)
- present and motivate all taken choices (Defend)
The grade for the course is computed as follows:
A retake requires at least a 4 for the original test.
- 25% Process (consistency, quality, and completeness of intermediate presentations)
- 25% Final project presentation
- 50% Final project report
To successfully complete this course, students should
- be fully fluent in coding in a mainstream programming language (C, C++, Python, Java, C#)
- be fully able to manage data/software on their own (platforms, tools, scripting, etcetera)
- have a strong background in math (linear algebra, calculus, optimization, statistics)
- have a background in visualization and/or graphics. Ideally, they will have taken a course that teaches general-level principles of visualization.
This course teaches Data Visualization methods with a focus on relational and high-dimensional data. Such data appear in many of real-life applications from science, social phenomena, and engineering.
The course is divided into two parts: relational and high-dimensional data, as follows.
We present a model used for representation and storing relational data, called graph or network, and list applications of network visualization.
We present methods that teach the visualization of large data tables (thousands of rows, hundreds of columns) from data science, artificial intelligence, and related fields.
We define so-called quality metrics to construct and assess network visualizations. In the following we study several network visualization methods, including:
- hierarchy (tree) visualization;
- visualizing general graphs.
- visualization of multilevel networks for modeling highly complex applications.
- graph bundling.
We discuss these methods as well as quality metrics to assess them including:
- parallel coordinate plots
- scatterplot matrices
- table lenses
- dimensionality reduction
The learning objectives are achieved through weekly lectures, group project work, and intermediate update meetings, as follows.
- are provided with datasets to visualize with the taught methods.
- implement presented methods and explain their design decisions.
- evaluate their visualizations with presented methods.
The course has no compulsory textbook. All needed information is given via slides, papers, and notes via the course’s website.