Professional Documents
Culture Documents
John F. Tripp
Introduction
• About me
• Assistant Professor at Clemson University (fourth year)
• PhD in IT Management from Michigan State University (2012)
• Teaching in the AMPBA since Batch #1.
• Over 17 years of “real-world” experience designing and developing rich user
interactive applications.
• I have 8 kids (25-9)
• Consulting
• LinkedIn (jftripp)
Overall Plan
• This weekend – theory and foundations.
• A little hands-on…
• Ongoing – hands-on applications of principles of visualization.
• Using Tableau.
• Yes, there are other software packages.
• No, we won’t be using them.
• This course will be concept heavy, and less focused on a particular software package.
• Any of the concepts that we learn in this course will be applicable on any platform.
Storytelling with Data
What we’re doing with Data Visualization…
• Fundamentally we are always attempting to do one of two things:
2. Provide the ability for a user to answer “their story” through a set of related
visualizations.
Does this tell a story?
Is it clear?
NO:
Many different,
unrelated data.
• Usually, a visualization can properly assist only one (or several highly
interrelated) evaluation(s) at a time.
Meta-Rules for Data Viz
1. The simplest chart is usually the one that communicates most clearly (e.g., use
the “not wrong” chart – not the “cool” chart)
2. Always DIRECTLY represent the relationship you are trying to communicate.
(e.g., don’t leave it to the viewer to calculate it from other information)
3. Never ask viewers to compare levels of a single dimension in two dimensions
(e.g., comparing differences in length is easy, comparing differences in area is
hard)
4. Never use color on top of color. (e.g., color is not absolute)
5. Do not violate the primal instincts of your viewers. (e.g., up means more)
6. Chart with graphical and ethical integrity! (e.g., don’t lie, either by mistake or
intentionally)
What is right/wrong here?
What is right/wrong here?
What is right/wrong here?
What is right/wrong here?
What is right/wrong here?
What is right/wrong here?
Meta-Rules for Data Viz
1. The simplest chart is usually the one that communicates most clearly (e.g., use
the “not wrong” chart – not the “cool” chart)
2. Always DIRECTLY represent the relationship you are trying to communicate.
(e.g., don’t leave it to the viewer to calculate it from other information)
3. Never ask viewers to compare levels of a single dimension in two dimensions
(e.g., comparing differences in length is easy, comparing differences in area is
hard)
4. Never use color on top of color. (e.g., color is not absolute)
5. Do not violate the primal instincts of your viewers. (e.g., up means more)
6. Chart with graphical and ethical integrity! (e.g., don’t lie, either by mistake or
intentionally)
Getting to the Heart of It
• Who?
• What?
• How?
Getting to the Heart of it…
Who is the audience?
• Do we always know?
• ASK
• Sit with them and watch them work. Ask them:
• “What are you thinking right now?”
• “What are you trying to answer at each step?”
• Do the work they are doing.
Getting to the Heart of it
• Step 1: Build the Cognitive Flow.
• Step 2: Identify the “similarity” or “relatedness” of each step of the
flow.
• Step 3: Identify the “most not wrong” visualization to use.
• Step 4: Identify the level of contextual assistance you must provide.
Step 1: Map out the COGNITIVE FLOW
• What’s the first question they need to answer…
• What’s the next question they need to answer…
• Etc.
• Etc.
• Etc.
• Are there BRANCHES that they pursue based upon the answers to any
question?
The Cognitive Flow
Is this an exploratory or
explanatory visualization
flow?
Step 2: Identify the ”Similarity” of the Steps
• When the questions are more similar, they lend themselves to being
presented on one screen.
• When they are more different, they should be split across multiple,
linked screens.
• We will spend time most weeks going over identifying the “most not
wrong” graphs for typical situations.
Step 4: Contextual Assistance
• Based upon several factors, various levels of contextual assistance are
required.
Method of Presentation
Individual Live
Method of Presentation
Individual Live
Step 4: Contextual Assistance
• Audience Contextual Knowledge
• Even when presenting live, you may need to give the audience additional
context.
• Annotations
• Comparisons
• Etc.
• When the user is working through the visualization themselves, you will need
to provide more context – regardless of audience knowledge.
• You may need to break out visualizations into additional “sub-steps”.
• Do NOT fall into the trap of trying to answer a highly complex question with an
“uber-graph” that tries to show all facets of the question.
• Complex questions usually require at least one dashboard plus additional graphs.