[Data Visualization] Justification

Review: Definitions of InfoVis

InfoVis is the use of computer-supported, interactive, visual representations of abstract data to amplify cognition. by Stuart Card, Jock Mackinlay, Ben Shneiderman, 1999

Questions We Will Answer

• Why Have a Human in the Loop?
• Why Have a Computer in the Loop?
• Why Use an External Representation?
• Why Depend on Vision?
• Why Show the Data in Detail?
• Why Use Interactivity?
• Why Is the Vis Idiom Design Space Huge?
• Why Focus on Tasks?
• Why Focus on Effectiveness?
• Why Are Most Designs Ineffective?
• Why Is Validation Difficult?
• Why Are There Resource Limitations?
• Why Analyze?

Why Have a Human in the Loop?

• If your job can be done fully automatically by a computer-based solution, there is no need for human judgement.
  • e.g., stock trading bot, number-plate recognition, …
• However, many analysis problems in practice are ill specified.
  • Even people don’t know how to approach the problem.
  • e.g., exploratory data analysis (EDA)
• There are many possible questions to ask, but people don’t know which of these questions are the right ones in advance.
• Start with a human in the loop.
  • Design a vis system to augment human capabilities rather than replacing the human in the loop completely.
• Through iterations, you will find important questions that should be answered to meet users’ need.
• If the questions can be answered automatically, e.g., by an ML solution, automate it.
• But, in most cases, the questions cannot be answered fully automatically and need a human in the loop.
  • Data, tasks, requirements are ever-changing in practice.

Why have a Computer in the Loop?

• You can draw a visualization either by hand with pencil and paper or by computerized drawing tools, for example, Illustrator.
• The scope of what people are willing and able to do manually is strongly limited by their attention span.
• Other limitations
  • Response time
  • Interactivity
  • Errors
  • Scalability
  • Generalizability

Why Use External Representation?

• External representations augment human capacity by allowing us to surpass the limitations of our own internal cognition and memory.
• Vis allows people to offload internal cognition and memory usage to the perceptual system, using carefully designed images as a form of external representations, sometimes also called external memory.
• What is 32 * 49?

Why Depend on Vision?

• Biggest bandwidth
  • Eyes (vision): 10,000,000 bps (1.25 MB/sec)
  • Skin (touch, tactition): 1,000,000 bps
  • Ears (hearing, audition): 100,000 bps
  • Nose (smell, olfaction): 100,000 bps
  • Mouth (taste, gustation): 1,000 bps
• Accurate
• Useful characteristics: preattentive processing
• Technological limitation

Why Show the Data in Detail?

• Vis tools help people when seeing the dataset structure in detail is better than seeing only a brief summary of it.
• Data exploration for finding patterns
• Assessing the validity of a statistical model

• Anscombe’s quartet and Simpson’s paradox

Why Use Interactivity?

• We cannot visualize all the dimensions of data in a single view, because the screen space is limited.
• We allow users to interactively choose what they want to see.
• Most modern visualization systems are interactive.
• Btw, in Card et al.’s definition of InfoVis, interactivity is a must.

Why is the Vis Idiom Space Huge?

• A vis idiom is a distinct approach to creating and manipulating visual representations.
  • e.g., a scatterplot is a vis idiom for bivariate data.
• If we only consider simple static idioms, such as scatterplots, bar charts, and line charts, there are not “too many” idioms.
• But, in practice, we link together these simple idioms and consider how to manipulate one or more idioms with interaction, the design space of possibilities gets even bigger.

Why Focus on Tasks?

• Which visualization is better?

• If you are aware of human-centered design, you must ask “for what?” first.
• A tool that serves well for one task can be poorly suited for another.

• In vis designs, we reframe the users’ task from domain-specific form into abstract form to identify similarities between tasks across many real-world usage contexts.
  • Wire transfer data
    • Task: find a transfer history between two bank accounts
  • Messenger data
    • Task: find a messaging record between two users

Why Focus on Effectiveness?

• Every year, many novel vis idioms are introduced in the visualization community.
• However, not all of them are effective.
• “It’s not just about making pretty pictures.”
• We want pictures that support user tasks.
• Any depiction of data is an abstraction where choices are made about which aspects to emphasize.
  • CAVEAT: You do not determine which aspects to emphasize as you like. The user tasks must be considered in the design process.

Why are Most Designs Ineffective?

• The vast majority of the possibilities in the design space will be ineffective for any specific usage context.
• Only a very small number of possibilities are in the set of reasonable choices, and of those only an even smaller fraction are excellent choices.
• Justification is important (why did you choose a specific idiom?).

Why is Validation Difficult?

• Because there are so many questions that you can ask.
• Fast?
• Insightful?
• Engaging?
• Familiar?
• Accurate?
• Satisfactory?
• Fun?

Why are there Resource Limitations?

• When you design or analyze a vis system, you must consider at least three different kinds of limitations:
• Computational capacity (e.g., scalability or responsiveness)
• CPU and RAM are finite!
• Human perceptual and cognitive capacity
• Human memory and attention are finite!
• Change blindness
• Display capacity
• # of pixels is finite!

Why are there Resource Limitations?

• Information density measures the amount of information encoded versus the amount of unused space.
  • a.k.a. graphic density and data-ink ratio
• Showing as much as possible at once vs showing too much at once

Why Analyze?

• In this lecture, we will analyze previous vis systems.
  • Many idioms and tools have been created in the past several decades.
  • There are so many possible combinations of data, tasks, and idioms.
  • We will abstract the previous vis systems!
• In terms of what? What, Why, and How
• What data does the user see? (data abstraction)
• Why does the user use the system? (task abstraction)
• How are the visual encoding and interaction idioms constructed? (idiomabstraction)
• One of these analysis trios is called an instance.

What-Why-How Examples