Evaluation of exercises/reports. The grading is based on an overall evaluation of exercises (50%) and final project report (50%). Specifically, the grade is based on individualized group reports.
Exam duration: Written exam: 4 hours
Evaluation: 7 step scale, internal examiner
General course objectives
The course objective is to enable students to create visualizations of complex data sets and to apply common strategies for understanding the content of data sets (e.g. text, music, images, etc).
Learning objectives
A student who has met the objectives of the course will be able to:
Access and assess types of available on-line data for data visualization. AI Concern: No big concerns.
Identify and use state-of-the-art tools to filter, clean, and organize large, complex datasets and argue why data integrity has not been violated. AI Concern: Mindlessly applying AI to solve the problem and not getting a sense of the underlying data.
Apply statistical tools to evaluate data visualization methods for exploration of single variable data, including dot and jitter plots, histograms, kernel density estimates, distribution functions, and more. AI Concern: One concern is that the students don’t learn to properly apply the statistical tools (that they don’t understand them) because the AI can solve the problems for them. The second one is that the AI can interpret the outputs for th e students.
Assess and apply data visualization methods for data exploration of multiple variable data, including estimating functional relationships (e.g. by smoothing noise, visualizing residuals, using log, semilog-plots, and simple regressions). AI Concern: That the AI could solve the “apply” part of the learning objective, thus causing the students to not learn the fundamental concepts.
Use visualization techniques to evaluate and identify limitations of summary statistics, based e.g. on Simpson’s paradox, and Anscombe’s quartet. AI Concern: Not a big concern. Although the worst students could still use AI to solve the issues.
Use basic principles of displaying visual information (e.g. Tufte’s six principles of graphical integrity) to create explanatory visualizations. AI Concern: No big concern. Although the worst students could still use AI to create Tufte graphics.
Apply specialized visualization software (e.g. JavaScript’s D3 library or Python libraries) in order to build custom visualizations designed to explain insights from a dataset to an audience. AI Concern: Here AI could be used directly. But maybe it’s OK. Explain insights is the important part here.
Analyze cases of narrative data visualization to extract the underlying principles used to construct this type of visualization. AI Concern: No big concerns.
Build a narrative data-visualization. AI Concern: No big concerns.
Content The course is based on mastering tools for analyzing data sets generated from online behavior. The course is structured around short lectures combined with exercises, as well as a high degree of independent project work.