How this course is structured, and how to follow along.
This course explores the role of visualization within data science workflows. We’ll get hands-on experience how visualization can be used to support,
The course mostly uses R, though occasionally we will use the vega-lite library in javascript. Prior experience with R, but not javascript, is expected.
In terms of content, we’ll begin with a three week crash course on the fundamental principles of data visualization. This will provide a vocabulary for designing and critiquing visualizations throughout the semester. Weeks 4 - 5 are focused on the connections between visualization and tidy data. Rearranging your data in the right way can make it much easier to visualize; conversely, a good interface can immediately highlight data quality issues.
From here on out, we’ll focus on exploratory analysis and model visualization. Weeks 6 - 8 consider data with temporal, geographic, or network structure – each comes with its own visual conventions. Weeks 9 - 11 focus on visualization of high-dimensional data, drawing from ideas in unsupervised learning. Weeks 12 and 13 describe the recent use of visualization to support inspection of complex supervised models. We close with a discussion about the possibilities of visualization in the broader intellectual landscape.
You can really only learn visualization by practicing. For this reason, most of the homeworks are technical exercises in building pre-defined visualizations (I do hope you can exercise more creativity through your course project, though). The homeworks will be much more approachable if you have already run all the code accompanying the recordings. I recommend you have a window open for running code while watching the recordings, pausing whenever you want to tinker with how a particular line works.
The .Rmd files discussed in each recording are linked at the top of the post. You should be able to run the .Rmd file directly on your computer – if you encounter any issues, don’t hesitate to reach out. Observable notebooks that reproduces figures discussed in the recordings are also linked at the top of each post. See the Introduction to Vega-Lite lecture for a brief description of how to reproduce examples in your own Observable notebooks.