The Week in Vis 11 (Mon, Nov 12 – Fri, Nov 16): Uncertainty and Modeling

by Mike Gleicher on November 9, 2018

Class Meetings
  • Mon, Nov 12 – Lecture:High-Dimensions
  • Wed, Nov 14 – Lecture:Modeling
  • Fri, Nov 16 – OPT:DC2 Demos
Week Deadlines

Last week, we talked about the challenges of scale. But we only really got to talk about reducing the number of items. We didn’t deal with the problems of having too many dimensions. And also a lot of DC2 work happened.

This week, we’ll spend more time talking about scale: starting with the problem of dealing with too many dimensions. The discussion of scale will lead naturally to the discussion of modeling and uncertainty: basically, models tend to be summaries.

One change for this week: I am going to try to better integrate ICEs into the regular lectures, since otherwise the lectures become too much of monologues.

Also, we will do DC2 demos – stay tuned for details. But basically, we’ll have class Friday (in 312, not 311) where we’ll give people the chance to show off their projects. Unfortunately, with a 60 student class, we cannot do all the demos in one class period – we’re still working out a plan.

You may want to look at this week’s learning goals Learning Goals 11: Week 11 – Uncertainty and Modeling.

Readings (due Mon, Nov 12 – preferably before class)

While uncertainty is a critical topic, there is no obvious good reading for it. Last year I gave a whole long list.

This year, I want you to read two:

  1. Boukhelifa, N., & Duke, D. J. (2009). Uncertainty visualization: why might it fail? In Proceedings of the 27th international conference extended abstracts on Human factors in computing systems – CHI EA ’09 (p. 4051). New York, New York, USA: ACM Press. doi:10.1145/1520340.1520616 (ACM – free access on campus or using UW library proxy).
  2. Correll, M., & Gleicher, M. (2014). Error Bars Considered Harmful: Exploring Alternate Encodings for Mean and Error. IEEE Transactions on Visualization and Computer Graphics, 20(12), 2142–2151. doi:10.1109/TVCG.2014.2346298 (web)

The point of uncertainty is that it means there is something in the data that is too hard for us to measure/model. Which makes a nice segway to considering modeling (since uncertainty usually involves a model), which is a hot topic because of that’s what Machine Learning is all about. This paper is particularly relevant if you are interested in the connection between Vis and Machine Learning (or Data Science more generally).

  1. M. Gleicher. A Framework for Considering Comprehensibility in Modeling. Big Data 4(2), June 2016. (page with PDF)

Optional

If you want to read more on uncertainty, I recommend:

  • Cumming, G., & Finch, S. (n.d.). Inference by eye: confidence intervals and how to read pictures of data. The American Psychologist, 60(2), 170–80. doi:10.1037/0003-066X.60.2.170 (pdf)

The statisticians have a lot to say about how we should think about uncertainty, especially in experiments. This paper gets at many of the issues (it is statisticians explaining to psychologists what they should do).

This one might be required in the future, but it’s too new for me to decide:
+ Jessica Hullman, Xiaoli Qiao, Michael Correll, Alex Kale, Matthew Kay. In Pursuit of Error: A Survey of Uncertainty Visualization Evaluation. IEEE Trans. Visualization & Comp. Graphics (Proc. InfoVis), 2019. (page with PDF)

Previous post:

Next post: