Skip to content

Latest commit

 

History

History
326 lines (204 loc) · 7.12 KB

lecture09.md

File metadata and controls

326 lines (204 loc) · 7.12 KB
Raw
title layout include_vega
Lecture 9
lecture
true

Data Visualization

## Matthew Turk ## Spring 2018 ## Lecture 9

Warm-Up Activity

  1. What is the visualization trying to show?
  2. What are its methods?
  3. What are the strengths / weaknesses?

"Previously on Data Viz..."

  • Basics of widgets and traitlets
  • Some bqplot
  • Network graphs

Today

  • Evaluating Visualization Systems
  • Assignment 3
  • Interactivity
  • More bqplot
  • Markdown (leading up to Idyll)

Evaluating Visualization Engines

  • Costs
  • Functionality
  • Aesthetics

Choices

  • Can I get ahold of this software?
  • Do I install it, or do I use it on a server?
  • What's the user interface like?
  • Is it declarative or is it procedural?

License: Software

  • What can you do with the software?
  • Can you study the software?
  • Who can you share it with?
  • Who can you give your derivative works to?

License: Software

License: Data

Accessibility

  • Is the software installed locally on your machine?
  • Is it hosted at a local or remote instance?
  • Who owns the visualizations, and how is access to them controlled?

Interface

How do you interact with the software?

  • Declarative: how do you want the plot to look?
  • Procedural: what are the steps to make the plot look that way?

Example Declarative

Chart(df).mark_bar().encode(
    X('precipitation', bin=True),
    Y('count(*):Q')
)

(From Altair Docs)

Evaluation: Costs

The "cost" of software is not exclusively the number of dollars you place on the counter when you get a big cardboard box with marketing blurbs on the side.

Think about cost in several ways:

  • Monetary cost for you to use the software
  • Monetary cost for someone else to view your creations
  • Temporal cost of setting up
  • Cognitive cost for learning and using the system
  • Transmission cost for sharing your creations

Evaluation: Aesthetics

Visualization is trendy.

When you construct something, think about the different ways it will be interpreted:

  • How will the viewer understand the story of the data?
  • What will the message of the visualization be?
  • Does the visualization say something about you and your handling of the data or utilization of tools?

Assignment 3

Your assignment is to pick three of the following possibilities and write up a set of comparisons for constructing the same visualization.

  • Matplotlib
  • Bokeh
  • Plotly
  • D3 / Vega / Vega-lite / Altair
  • bqplot
  • R / RStudio

These comparisons should be:

  • What is the license for the software?
  • What is the focus of the software?
  • Does it have interactivity, and how easy is it?
  • What are the pros and cons of using it?

Interactivity

This week, we'll talk about some basics principles of interactivity in visualization.

What do you think of when you think of interactive visualizations?

Interactivity: Parameters

  • Point characteristics
  • Axis limits/bounds
  • Transform/scale

Interactivity: Parameters

  • Click-and-drag
  • Rectangle zoom
  • Adjustment

Interactivity: Linking & Brushing

Interactivity: Linking & Brushing

filter( variable2 > variable1 )

Interactivity: Linking & Brushing

filter( variable2 > variable1 )

Interactivity: Linking & Brushing

filter( variable2 > variable1 )

Interactivity: Linking & Brushing

Interactivity: Linking & Brushing

Interactivity: Linking & Brushing

<iframe width="1024" height="576" src="https://www.youtube.com/embed/B7XoW2qiFUA" frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>

Implementing This

Two main approaches to the selection process:

  • Concurrent filtering
  • Index-based selection

What are the pros and cons of each?

What are methods of showing "linked" and "brushed" data if you have:

  • Scatter plot
  • Histogram
  • Field / image plot

More bqplot

Today, we return to bqplot to explore interactivity.

Recall that with bqplot, we construct a set of objects that are related:

  • Scales
  • Axes
  • Marks
  • Figures
  • Interactions

bqplot interaction

As noted in a previous class, bqplot widgets are all based on ipywidgets. This means we use the same systems for describing the two.

We add an interaction to a given figure via the interaction keyword argument to a figure.

bqplot interactors

We will be able to use these different interaction methods:

  • FastIntervalSelector
  • IndexSelector
  • BrushIntervalSelector & BrushSelector
  • MultiSelector
  • LassoSelector
  • HandDraw
  • PanZoom
  • Tooltip

Markdown

Next week, we will be discussing vega-lite in detail, as well as idyll.

We will be using markdown. You can think of it as standard text that we mark "up" to provide indications of formatting:

  • To bold, use **two asterisks**
  • You can italicize with _underscores_
  • Try out making bullet lists with * on each line
  • Links are fun: [link text](http://example.com/)