Skip to content

Instantly share code, notes, and snippets.

@caseywdunn
Last active April 5, 2021 18:29
Show Gist options
  • Save caseywdunn/6c8b9f0d2117fc11bee4580807c34448 to your computer and use it in GitHub Desktop.
Save caseywdunn/6c8b9f0d2117fc11bee4580807c34448 to your computer and use it in GitHub Desktop.

Data Visualization

Some basic notes for a couple hour discussion about data visualization.

Build on a few points from my previous course.

To read in advance of the class: https://r4ds.had.co.nz/data-visualisation.html .

Agenda

The goal

Data visualization is the mapping of data to aesthetic properties

What are data? Most data are measurements. Measurement is "the assignment of numbers to attributes of the natural world" (Houle et al.). Measurement theory is a rich body of work that explains what this actually means. I encourage you to check it out.

What are aesthetic properties? This is the domain of art, visual design, and psychology. In this context, we will consider aesthetic properties to be attributes we can change that lead to a discernable visual difference. They include things such as:

  • Color
  • X position
  • Y position
  • Shape
  • Transparency
  • Texture
  • Size

This domain of attributes is often described within the context of the Principles of Design.

So data visualization never exists on its own. It is always part of a workflow, which often looks something like this:

Natural world --(measurement)--> 
Data --(map to aesthetic attributes)--> 
Visualization --(perception and cognition)--> 
Insight | feeling | action

These relationships are transitive. Because insight tracks the visualization which tracks the data which tracks the natural world, insight tracks the natural world. In effective data visualization, this tracking will be as clear is possible.

Effective practice

As a practioner of data vizualization, you are making a series of decisions about the mapping of aesthetic attributes to data (and ultimately the natural world). These include:

  • Which features of the data to not include in the vizualization
  • Which features of the data to include
  • Which aesthetic properties to map them to
  • Which aesthetic attributes to add that don't map to data

Visualizations almost always are extreme simplifications of the original natural world - much is left out. This is a feature, not a bug.

Sometimes visualization closely resembles the natural world, such as when points are placed on a map.

Sometimes visualization greatly alters the way the world is seen, such as when petal lengths are mapped to x position.

Sometimes visuzlizations show things that have no visual representation in the natural world, such as charges of particles. Then the workflow is mapping our natural senses to new senses, and expanding our power to perceive the universe.

Visualizaiton examples

Present visualizations:

  • Explicitly describe which data features are mapped to which aesthetic attributes
  • Which aesthetic attributes don't map to data
  • Is it effective?
  • Anything you would have dine differently?

Visualization exercises

Walk through https://r4ds.had.co.nz/data-visualisation.html

Discussion

How is this landscape changing as we use new technologies to communicate? As communication needs change?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment