Your Presenter
Greg Nelson, CPHIMS, MMCi
Vice President, Analytics & Strategy – Vidant Health
• Recovering Social Psychologist turned Technologist/ Data
Scientist/ Informaticist (Duke University)
• Prolific writer/ presenter (175+ publications/ presentations)
• 25+ year Analytics/ Data Science Expert
• Adjunct Faculty Fuqua School of Business (Adv Analytics &
Data Visualization)
• Passionate about developing the next generation of data
champions
• Author: The Analytics Lifecycle Toolkit (Wiley, 2018)
Data Viz Best Practices
Storytelling for Good:
Data Visualization Best Practices
Greg Nelson, MMCi, CPHIMS
Vice President, Analytics & Strategy
Vidant Health
@gregorysnelson
Activity
PARTICIPANTS Class.
TIME 5 minutes.
WHAT YOU NEED Pen and paper.
On that paper, divide the page into 6
rows and …4 columns.
Label the rows 1 through 6 and the
columns A through D.
PROMPT
You will be presented a series of graphical displays.
Rate each of them on the following scale
1: Definitely not
2: Somewhat not
3: In between
4: Somewhat
5: Definitely
The graphic was:
A … easy to understand
B … interesting
C … surprising
CONSIDER: Jot a quick note that answers this question:
D … What story was being told?
Data Viz Best Practices
1. Twitter mentions
Data Viz Best Practices
2. Basketball shots
Data Viz Best Practices
3. Three-point shots
Data Viz Best Practices
4. Fish population
Data Viz Best Practices
5. Cancer deaths
Data Viz Best PracticesData Viz Best Practices
6. Healthcare quality
Data Viz Best Practices
Data stories.. What did we learn?
• Which one was interesting?
• Which one was surprising?
• Which was easy to understand?
• Which story was most
compelling (influence to action)?
1
4
2 3
6
5
Data Viz Best Practices
If you can find and understand the
root cause of their [emotions], you
can turn data into a story that will
stay with your audience, conveying
exactly what you are trying to
communicate and solve.
Andy Kirk, Helen Kennedy & Jeremy
Boy
Stories can move, mobilize, and motivate people toward change and action…
vSource and Perception
vTime
vSubject Matter
vContext
vSkill Level Confidence
Andy Kirk, Helen Kennedy & Jeremy Boy
What factors affect the data visualization consumption and engagement process?
Data Viz Best Practices
How process information
Data Viz Best Practices
• Learn how to critically evaluate data
visualizations
• Identify graphical techniques that are often
used to mislead an audience
• Understand the competencies that are
important in visual storytelling
• Gain an appreciation for how design
thinking can aid in crafting a compelling
data story
• Quickly spot data visualizations that don’t
tell the entire story
Learning objectives…
Benefits of Data Visualization
• Great visualizations are efficient
• Visualizations can help you achieve more insight
and understanding
• Can help create a shared view and alignment on
actions
Adapted from Swimming in Data? Three Benefits of Visualization John Sviokla – HBR December, 2009
Data Viz Best Practices
Predictive outcome
for selected
medication
Patient
demographics
Profile of outcome
response to
prescribed medications
Profile of about
prescribed medications
and therapy
Treatment evidence
aggregated from
comparative population
Button to open
filter panel
Health Outcomes Analysis
Source: Ketan Mane, PhD, Senior Research Scientist at RENCI, University of North Carolina at Chapel Hill
Data Viz Best Practices
Types of Data Visualizations
Data Visualization
• Parallel coordinates
• Social network
• Arc
• Matrix views
• Node-link
• Geospatial
• Heat maps
• Bubble charts
• Spark
• Spider
• Word clouds
Traditional “Stat Graphics”
• Time series (line, multiline, area)
• Magnitude (column, bar, pie, donut)
• Stacked (area, bar)
• Small multiples (spark, bullet)
• Horizon graphs
• Statistical distributions
• Maps (choropleth, symbols)
• Multidimensional (principal
components, hierarchical clustering,
structural equation models)
• Association (forest plots, regression
trees, scatter)
Mind maps
Adjacency diagrams
Enclosure diagrams
Elastic lists
Graphic similarity models
infosthetics / dataesthetics
Pictographs
Data Viz Best Practices
Analytics
competency relates
to the knowledge,
skills, abilities and
disposition required
to successfully turn
data into actionable
interventions.
Top 10 Competencies
1. Data storytelling
2. Question design
3. Requirements Management
4. Business Impact Assessment
5. Data visualization Techniques
6. Tool agility and technical fluency
7. Statistical principles
8. Profiling and Characterization
9. Data-driven Decision Making
10. Statistical literacy
Proficiency in storytelling for actionGregory S. Nelson
The Analytics Lifecycle Toolkit, 2018
Visual Display of Data
• What is the difference between
those who design visual displays
versus other types of “data
junkies”?
• What tools are appropriate for
each?
Source: Juice Analytics
Data Viz Best Practices
Our challenge…
From…. To…
Data Viz Best Practices
Data Visualization Roles
Data Viz Best Practices
Data author
Storyteller
Data
consumer
Adapted from Data Fluency (Juice Analytics)
Hypothesis- or Data-Driven?
HISTORICAL VIEW
Theoretical
Framework
Testable
Hypothesis
Data
Investigation
Empirical Study
Data
Investigation
Testable
Hypothesis
Empirical Study
Theoretical
Framework
ALTERNATE VIEW
Data Viz Best Practices
General ”Analytics” Competencies by Task
Business / Domain
Knowledge
Technical Skills
(Computer science,
Technology, Programming)
Math/ Statistics
v Formulate a question or
problem statement ❉
v Generate a hypothesis
that is testable ❉
v Gather/ generate data
to understand the
phenomenon
❉ ❉ ❉
v Analyze data to test the
hypothesis/ draw
conclusions
❉ ❉
v Communicate results to
interested parties or
take action
❉ ❉
Data Viz Best Practices
Data Viz Best Practices
DEFINE
Stakeholder
analysis
Requirements
gathering &
elicitation
Problem definition
Question design
Expected benefit
EXPLORE
Exploration of data (breadth &
depth)
Data visualization (explore)
Identification of data
relationships
Documentation of dataset culture
Generation of descriptive
statistics
IDENTIFY
Data extraction
Data integration
Data transformation
ANALYZE
Statistical analysis
Hypothesis testing
Enrichment options
Modeling
PRESENT
Data visualization (inform)
Storyboarding
Results presentation
ROI calculation
Documentation
OPERATIONALIZE
Workflow impact
End-user training
Analytic product calibration
Maintenance
Retuning and improvement
Analytics Product Lifecycle Management
Data Viz Best Practices
The goals of data visualization
Data Viz Best Practices
“A graph’s primary purpose is to describe and communicate the shapes that
represent properties of and relationships among quantitative variables.”
Nathan Yau categorizes visualization goals in these terms :
• Patterns
• Proportions
• Relationships
• Comparisons
Data Visualization
Catalogue
A Periodic Table Of
Visualization Methods
Communication
Understanding
Action
Data Viz Best Practices
What we visualize
Data Viz Best Practices
Visualizations
Whole vs. Part
Simple Comparison
Multi Comparison
Trends
Frequencies
Correlation/ Relationship/
Proportions
Spatial Relationships
http://www.datavizcatalogue.com/search.html
Design Matters?
• Design Best Practices
• Readable
• Clear
• Unambiguous
Data Viz Best Practices
Visual
design and
visual cues
Placement,
Proximity and
Position
Lines, shapes
and colors
Coordinate
systems
Measurement
scales
Visual
hierarchy
Data Viz Modalities
Data Viz Best Practices
Visual AnalyticsMaps
Charts
Graphs
Infographics
Dashboards
Interactions
Mobile BI
Animations
Motion Infographic
Data Journalism
Exploration/ Discovery
Consume Explore
Art or Craft?
Artist Engineer
God and Moses? The Differences Between Edward Tufte and Stephen Few
Data Viz Best Practices
Art or Craft?
Artist Engineer
Data Viz Best Practices
Analysis “Gotchas”
Methodological
Statistical Analysis
Interpretation and
Communication
Results
Operationalization
Actionability
Thinking and Intelligence
Cognitive Biases
Data Viz Best Practices
Common Data Visualization Mistakes
1 Unfair comparisons between two or more elements
when the scale reflects only part of the whole (e.g.,
cropped axes.) When numbers don’t add up, scales
don’t make sense or the arrangements are counter-
intuitive, we lose credibility.
2
3
4
5
6
7
8
Improper Scales
In graphs with multiple axes, people can often make
correlations based on the trends even though the
scales are unrelated.
Apples to Oranges Comparisons
A visual association doesn’t necessarily mean that
one thing causes another. The form of the change
isn’t necessarily the cause of the change.
Implying Causation
Ignoring population size can make an effect seem
much more dramatic than it really is.
Understanding Adjustments
Choosing the wrong format can devastating to your
story. Similarly issues can arise when you try to do
too much or try to oversimplify or focus on the
“pretty”.
Chart Junk
Don’t omit key variables and make sure that all
relevant data is presented.
Incomplete Data
The visual is only part of the narrative. Don’t feel like
you can’t augment the visual display with relevant
information that rounds out the narrative.
Not using Annotations
When we present the data in only one way, we limit
the ability to explore and create connections and
associations that we may not have considered.
Presenting the data in multiple ways helps people
understand the whole picture and one
representation may resonate more than another.
Incomplete story
Data Viz Best Practices
Improper Scales
Broken scales show drama where it doesn't exist.
http://news.nationalgeographic.com/2015/06/150619-data-points-five-ways-to-lie-with-charts/
Data Viz Best Practices
Improper Scales
Numbers do not add up to 100%
Data Viz Best Practices
Apples to Oranges Comparisons
Data Viz Best Practices
http://news.nationalgeographic.com/2015/06/150619-data-points-five-ways-to-lie-with-charts/
Greg‘s Data Viz Principles
Helps…
• Process lots of “data”
• Aids in understanding
(comprehension)
• Contextualizes the story
• Focuses your attention on the story
• Reduces complexity
• Makes us think
• Allows for exploration and self-guided
discovery
• Helps to create a shared sense of what
should happen (leads to action)
Hurts…
• Tells us only the part they want us
to know
• Dumb down the story
• Confuses us
• Is just pretty
• Too technical
Data Viz Best Practices
DEVELOPING YOUR DATA STORY
The Big Idea
Design for
Action
Prototype Activate
Data Viz Best Practices
Pixar’s Rules of Storytelling
Data Viz Best Practices
#2: You gotta keep
in mind what's
interesting to you as
an audience, not
what's fun to do as
a writer. They can
be very different.
Data Viz Best Practices
#7: Come up with
your ending before
you figure out your
middle. Seriously.
Endings are hard,
get yours working
up front.
Data Viz Best Practices
#8: Finish your
story, let go even if
it's not perfect. In
an ideal world you
have both, but
move on. Do better
next time.
Data Viz Best Practices
#11: Putting it on
paper lets you start
fixing it. If it stays in
your head, a perfect
idea, you'll never
share it with
anyone.
Data Viz Best Practices
#14: Why must you
tell THIS story?
What's the belief
burning within you
that your story
feeds off of? That's
the heart of it.
Data Viz Best Practices
#22: What's the
essence of your
story? Most
economical telling
of it? If you know
that, you can build
out from there.
Data Viz Best Practices
Question and Answers
@gregorysnelson
linkedin.com/in/gregorysnelson
greg.nelson@vidanthealth.com
919.931.4736
Contact

Data is love data viz best practices

  • 1.
    Your Presenter Greg Nelson,CPHIMS, MMCi Vice President, Analytics & Strategy – Vidant Health • Recovering Social Psychologist turned Technologist/ Data Scientist/ Informaticist (Duke University) • Prolific writer/ presenter (175+ publications/ presentations) • 25+ year Analytics/ Data Science Expert • Adjunct Faculty Fuqua School of Business (Adv Analytics & Data Visualization) • Passionate about developing the next generation of data champions • Author: The Analytics Lifecycle Toolkit (Wiley, 2018) Data Viz Best Practices
  • 2.
    Storytelling for Good: DataVisualization Best Practices Greg Nelson, MMCi, CPHIMS Vice President, Analytics & Strategy Vidant Health @gregorysnelson
  • 3.
    Activity PARTICIPANTS Class. TIME 5minutes. WHAT YOU NEED Pen and paper. On that paper, divide the page into 6 rows and …4 columns. Label the rows 1 through 6 and the columns A through D. PROMPT You will be presented a series of graphical displays. Rate each of them on the following scale 1: Definitely not 2: Somewhat not 3: In between 4: Somewhat 5: Definitely The graphic was: A … easy to understand B … interesting C … surprising CONSIDER: Jot a quick note that answers this question: D … What story was being told? Data Viz Best Practices
  • 4.
    1. Twitter mentions DataViz Best Practices
  • 5.
    2. Basketball shots DataViz Best Practices
  • 6.
    3. Three-point shots DataViz Best Practices
  • 7.
    4. Fish population DataViz Best Practices
  • 8.
    5. Cancer deaths DataViz Best PracticesData Viz Best Practices
  • 9.
    6. Healthcare quality DataViz Best Practices
  • 10.
    Data stories.. Whatdid we learn? • Which one was interesting? • Which one was surprising? • Which was easy to understand? • Which story was most compelling (influence to action)? 1 4 2 3 6 5 Data Viz Best Practices
  • 11.
    If you canfind and understand the root cause of their [emotions], you can turn data into a story that will stay with your audience, conveying exactly what you are trying to communicate and solve. Andy Kirk, Helen Kennedy & Jeremy Boy
  • 12.
    Stories can move,mobilize, and motivate people toward change and action… vSource and Perception vTime vSubject Matter vContext vSkill Level Confidence Andy Kirk, Helen Kennedy & Jeremy Boy What factors affect the data visualization consumption and engagement process? Data Viz Best Practices
  • 13.
    How process information DataViz Best Practices
  • 14.
    • Learn howto critically evaluate data visualizations • Identify graphical techniques that are often used to mislead an audience • Understand the competencies that are important in visual storytelling • Gain an appreciation for how design thinking can aid in crafting a compelling data story • Quickly spot data visualizations that don’t tell the entire story Learning objectives…
  • 15.
    Benefits of DataVisualization • Great visualizations are efficient • Visualizations can help you achieve more insight and understanding • Can help create a shared view and alignment on actions Adapted from Swimming in Data? Three Benefits of Visualization John Sviokla – HBR December, 2009 Data Viz Best Practices
  • 16.
    Predictive outcome for selected medication Patient demographics Profileof outcome response to prescribed medications Profile of about prescribed medications and therapy Treatment evidence aggregated from comparative population Button to open filter panel Health Outcomes Analysis Source: Ketan Mane, PhD, Senior Research Scientist at RENCI, University of North Carolina at Chapel Hill Data Viz Best Practices
  • 17.
    Types of DataVisualizations Data Visualization • Parallel coordinates • Social network • Arc • Matrix views • Node-link • Geospatial • Heat maps • Bubble charts • Spark • Spider • Word clouds Traditional “Stat Graphics” • Time series (line, multiline, area) • Magnitude (column, bar, pie, donut) • Stacked (area, bar) • Small multiples (spark, bullet) • Horizon graphs • Statistical distributions • Maps (choropleth, symbols) • Multidimensional (principal components, hierarchical clustering, structural equation models) • Association (forest plots, regression trees, scatter) Mind maps Adjacency diagrams Enclosure diagrams Elastic lists Graphic similarity models infosthetics / dataesthetics Pictographs Data Viz Best Practices
  • 18.
    Analytics competency relates to theknowledge, skills, abilities and disposition required to successfully turn data into actionable interventions. Top 10 Competencies 1. Data storytelling 2. Question design 3. Requirements Management 4. Business Impact Assessment 5. Data visualization Techniques 6. Tool agility and technical fluency 7. Statistical principles 8. Profiling and Characterization 9. Data-driven Decision Making 10. Statistical literacy Proficiency in storytelling for actionGregory S. Nelson The Analytics Lifecycle Toolkit, 2018
  • 19.
    Visual Display ofData • What is the difference between those who design visual displays versus other types of “data junkies”? • What tools are appropriate for each? Source: Juice Analytics Data Viz Best Practices
  • 20.
  • 21.
    Data Visualization Roles DataViz Best Practices Data author Storyteller Data consumer Adapted from Data Fluency (Juice Analytics)
  • 22.
    Hypothesis- or Data-Driven? HISTORICALVIEW Theoretical Framework Testable Hypothesis Data Investigation Empirical Study Data Investigation Testable Hypothesis Empirical Study Theoretical Framework ALTERNATE VIEW Data Viz Best Practices
  • 23.
    General ”Analytics” Competenciesby Task Business / Domain Knowledge Technical Skills (Computer science, Technology, Programming) Math/ Statistics v Formulate a question or problem statement ❉ v Generate a hypothesis that is testable ❉ v Gather/ generate data to understand the phenomenon ❉ ❉ ❉ v Analyze data to test the hypothesis/ draw conclusions ❉ ❉ v Communicate results to interested parties or take action ❉ ❉ Data Viz Best Practices
  • 24.
    Data Viz BestPractices
  • 25.
    DEFINE Stakeholder analysis Requirements gathering & elicitation Problem definition Questiondesign Expected benefit EXPLORE Exploration of data (breadth & depth) Data visualization (explore) Identification of data relationships Documentation of dataset culture Generation of descriptive statistics IDENTIFY Data extraction Data integration Data transformation ANALYZE Statistical analysis Hypothesis testing Enrichment options Modeling PRESENT Data visualization (inform) Storyboarding Results presentation ROI calculation Documentation OPERATIONALIZE Workflow impact End-user training Analytic product calibration Maintenance Retuning and improvement Analytics Product Lifecycle Management Data Viz Best Practices
  • 26.
    The goals ofdata visualization Data Viz Best Practices “A graph’s primary purpose is to describe and communicate the shapes that represent properties of and relationships among quantitative variables.” Nathan Yau categorizes visualization goals in these terms : • Patterns • Proportions • Relationships • Comparisons Data Visualization Catalogue A Periodic Table Of Visualization Methods
  • 27.
  • 28.
    What we visualize DataViz Best Practices Visualizations Whole vs. Part Simple Comparison Multi Comparison Trends Frequencies Correlation/ Relationship/ Proportions Spatial Relationships http://www.datavizcatalogue.com/search.html
  • 29.
    Design Matters? • DesignBest Practices • Readable • Clear • Unambiguous Data Viz Best Practices Visual design and visual cues Placement, Proximity and Position Lines, shapes and colors Coordinate systems Measurement scales Visual hierarchy
  • 30.
    Data Viz Modalities DataViz Best Practices Visual AnalyticsMaps Charts Graphs Infographics Dashboards Interactions Mobile BI Animations Motion Infographic Data Journalism Exploration/ Discovery Consume Explore
  • 31.
    Art or Craft? ArtistEngineer God and Moses? The Differences Between Edward Tufte and Stephen Few Data Viz Best Practices
  • 32.
    Art or Craft? ArtistEngineer Data Viz Best Practices
  • 33.
    Analysis “Gotchas” Methodological Statistical Analysis Interpretationand Communication Results Operationalization Actionability Thinking and Intelligence Cognitive Biases Data Viz Best Practices
  • 34.
    Common Data VisualizationMistakes 1 Unfair comparisons between two or more elements when the scale reflects only part of the whole (e.g., cropped axes.) When numbers don’t add up, scales don’t make sense or the arrangements are counter- intuitive, we lose credibility. 2 3 4 5 6 7 8 Improper Scales In graphs with multiple axes, people can often make correlations based on the trends even though the scales are unrelated. Apples to Oranges Comparisons A visual association doesn’t necessarily mean that one thing causes another. The form of the change isn’t necessarily the cause of the change. Implying Causation Ignoring population size can make an effect seem much more dramatic than it really is. Understanding Adjustments Choosing the wrong format can devastating to your story. Similarly issues can arise when you try to do too much or try to oversimplify or focus on the “pretty”. Chart Junk Don’t omit key variables and make sure that all relevant data is presented. Incomplete Data The visual is only part of the narrative. Don’t feel like you can’t augment the visual display with relevant information that rounds out the narrative. Not using Annotations When we present the data in only one way, we limit the ability to explore and create connections and associations that we may not have considered. Presenting the data in multiple ways helps people understand the whole picture and one representation may resonate more than another. Incomplete story Data Viz Best Practices
  • 35.
    Improper Scales Broken scalesshow drama where it doesn't exist. http://news.nationalgeographic.com/2015/06/150619-data-points-five-ways-to-lie-with-charts/ Data Viz Best Practices
  • 36.
    Improper Scales Numbers donot add up to 100% Data Viz Best Practices
  • 37.
    Apples to OrangesComparisons Data Viz Best Practices http://news.nationalgeographic.com/2015/06/150619-data-points-five-ways-to-lie-with-charts/
  • 38.
    Greg‘s Data VizPrinciples Helps… • Process lots of “data” • Aids in understanding (comprehension) • Contextualizes the story • Focuses your attention on the story • Reduces complexity • Makes us think • Allows for exploration and self-guided discovery • Helps to create a shared sense of what should happen (leads to action) Hurts… • Tells us only the part they want us to know • Dumb down the story • Confuses us • Is just pretty • Too technical Data Viz Best Practices
  • 39.
    DEVELOPING YOUR DATASTORY The Big Idea Design for Action Prototype Activate Data Viz Best Practices
  • 40.
    Pixar’s Rules ofStorytelling Data Viz Best Practices
  • 41.
    #2: You gottakeep in mind what's interesting to you as an audience, not what's fun to do as a writer. They can be very different. Data Viz Best Practices
  • 42.
    #7: Come upwith your ending before you figure out your middle. Seriously. Endings are hard, get yours working up front. Data Viz Best Practices
  • 43.
    #8: Finish your story,let go even if it's not perfect. In an ideal world you have both, but move on. Do better next time. Data Viz Best Practices
  • 44.
    #11: Putting iton paper lets you start fixing it. If it stays in your head, a perfect idea, you'll never share it with anyone. Data Viz Best Practices
  • 45.
    #14: Why mustyou tell THIS story? What's the belief burning within you that your story feeds off of? That's the heart of it. Data Viz Best Practices
  • 46.
    #22: What's the essenceof your story? Most economical telling of it? If you know that, you can build out from there. Data Viz Best Practices
  • 47.
  • 48.