Visualization for Machine Learning Week 2 Lab Recap

Visualization for Machine Learning Week 2 Lab Recap

Miscellaneous Notes

The slides I showed this week can be found here.

Any questions about what we cover during the lab can be directed to my email, erin.mcgowan@nyu.edu.

My office hours are TBD, and will be held online (by popular vote, as opposed to in person on the Brooklyn campus). You can log your availability at the office hours poll here. I’ll try to select a time when most people are available.

Please join the Discord if you haven’t already! This is where all class communication will take place.

Gestalt Principles

We introduced the Gestalt Principles by watching the video found here. We then discussed the rules of good figure, proximity, similarity, continuation, closure, and symmetry. These principles are important to keep in mind while creating data visualizations, as they will help you anticipate how a viewer may interpret your visualization.

Data Types, Graphical Marks, and Visual Encoding Channels

We walked through Jeffrey Heer’s Data Types, Graphical Marks, and Visual Encoding Channels notebook this week. Specifically, we discussed the difference between nominal, ordinal, quantitative, and temporal data types, and explored how to approach visualizing each with different graphical marks and visual encoding channels on an example scatterplot. We also continued our discussion of anticipating how the viewer may interpret (or misinterpret) your visualization, and aligning the visualization with the viewer’s intuition where possible. For instance, we saw that representing higher population with larger dots in our scatterplot and lower population with smaller marks made more sense intuitively than representing lower population with larger dots and higher population with smaller dots.

Other Resources

The NYUVIS Guides and Examples page is extremely helpful for all things D3.

Note: Don’t worry about absorbing all of these immediately! We will dive into more complex and interactive visualizations together as the semester continues, but this is a useful reference if you want to create a visualization but aren’t sure where to start.

However, if you are new to Javascript, I would recommend going over the Javascript Basics notebook this week. If you are new to D3, I recommend taking a look at the Data Transformation and SVG and D3 Basics notebooks - we covered some of these functions during the lab this week, but these go into more detail and introduce a few other simple but helpful functions. Building a strong understanding of the basics now will help you in future labs.