INFOVIS8803DV > SPRING 17
OVERVIEW OF DATA VISUALIZATIONVisual Representations + Interaction Techniques
Prof. Rahul C. Basole
CS/MGT 8803-DV > January 11, 2017
INFOVIS8803DV > SPRING 17
Administrative
• Course Website
– Syllabus
– Schedule
– Assignments
– Grading
– Instructor & TA
– DataVis Resources
• T-Square
• Piazza– piazza.com/gatech/spring2017/8803
• Tumblr
datavis17.wordpress.com
Please cc TA on emails
INFOVIS8803DV > SPRING 17
Name Cards?
INFOVIS8803DV > SPRING 17
Recall: What is DataVis?
• DataVis is presenting data via interactive charts, graphs, maps so
that users can understand the data, answer questions about the
data and gain insights from the data
• It’s REPRESENTATION + INTERACTION
INFOVIS8803DV > SPRING 17
DataVis ≠ Scientific Visualization (SciVis) & Medical Data Visualization (MedVis)
INFOVIS8803DV > SPRING 17
DataVis ≠ SciVis & MedVis (cont.)
• SciVis/MedVis Data generally associated with physical positions in a
2D or 3D space – data has some “geometry”
• In DataVis, the data is generally abstract. We have to create a
geometry with which the data is encoded.
INFOVIS8803DV > SPRING 17
Biological Visualization (BioVis)
INFOVIS8803DV > SPRING 17
Why do we have humans in the
decision making loop?
INFOVIS8803DV > SPRING 17
Why have a computer in the loop?
INFOVIS8803DV > SPRING 17
And why is DataVis so important today?
INFOVIS8803DV > SPRING 17
Data Overload
INFOVIS8803DV > SPRING 17
Data Overload (cont.)
INFOVIS8803DV > SPRING 17
INFOVIS8803DV > SPRING 17
Data Overload (cont.)
• How to make use of data?
• How do we make sense of data?
• How do we harness data in decision-making processes?
• How do we avoid being overwhelmed?
INFOVIS8803DV > SPRING 17
The Challenge
Transform the data into information (understanding,
sensemaking, insight) thus making it useful to people.
INFOVIS8803DV > SPRING 17
The Problem
How?
Data Transfer
Web,
Books,
Papers,
Game Scores,
Scientific Data,
Biotech,
Shopping,
People,
Stock/Finance,
Social Media,
News,
…
Vision: 100 MB/s
Ears: <100 b/s
Telepathy
Haptic/Tactile
Smell
Taste
(BIG, WIDE, SMALL)
DATA
INFOVIS8803DV > SPRING 17
Human Vision
• Highest bandwidth sense
• Fast, parallel
• Pattern recognition
• Pre-attentive
• Extends memory and
cognitive capacity
• People think visually
Impressive.
Let’s use it!
INFOVIS8803DV > SPRING 17
INFOVIS8803DV > SPRING 17
INFOVIS8803DV > SPRING 17
1 2 8 1 7 6 8 7 5 6 1 3 8 9 7 6 5 4 6 9 8 4 5 0 6 9 8 5 6 0 4 9 8 2 8 2 6 7 6 2
9 8 0 9 8 5 8 4 5 8 2 2 4 5 0 9 8 5 6 4 5 8 9 4 5 0 9 8 4 5 0 9 8 0 9 4 3 5 8 5
9 0 9 1 0 3 0 2 0 9 9 0 5 9 5 9 5 9 5 7 7 2 5 6 4 6 7 5 0 5 0 6 7 8 9 0 4 5 6 7
3 8 4 5 7 8 9 8 0 9 8 2 1 6 7 7 6 5 4 8 7 6 3 6 4 9 0 8 5 6 0 9 1 2 9 4 9 6 8 6
INFOVIS8803DV > SPRING 17
1 2 8 1 7 6 8 7 5 6 1 3 8 9 7 6 5 4 6 9 8 4 5 0 6 9 8 5 6 0 4 9 8 2 8 2 6 7 6 2
9 8 0 9 8 5 8 4 5 8 2 2 4 5 0 9 8 5 6 4 5 8 9 4 5 0 9 8 4 5 0 9 8 0 9 4 3 5 8 5
9 0 9 1 0 3 0 2 0 9 9 0 5 9 5 9 5 9 5 7 7 2 5 6 4 6 7 5 0 5 0 6 7 8 9 0 4 5 6 7
3 8 4 5 7 8 9 8 0 9 8 2 1 6 7 7 6 5 4 8 7 6 3 6 4 9 0 8 5 6 0 9 1 2 9 4 9 6 8 6
INFOVIS8803DV > SPRING 17
Consider a (classical) example:
The Anscombe’s Quartet
• Statistics are the same for each of the four x-y tables (the quartet)
– Mean = 9
– Variance = 9
– Correlation = 0.816
– Linear regression: y = 3 + 0.5x
• So what’s different about the data?
• You could study the tables very closely (and make little progress)
• Or you could visualize the data!
F.J. Anscombe, “Graphs in Statistical Analysis”, American Statistician, February 1973, 17-21.
INFOVIS8803DV > SPRING 17
Take out Paper + Pencil!
INFOVIS8803DV > SPRING 17
HIDE
INFOVIS8803DV > SPRING 17
Another example … Which cereal has the most/least potassium?
Is there a relationship between potassium and fiber?
If so, are there any outliers?
Which manufacturer makes the healthiest cereals?
Questions:
INFOVIS8803DV > SPRING 17
INFOVIS8803DV > SPRING 17
Visualization
• Often thought of as process of making a graphic or an image
• Really is a cognitive process
– Form a mental image of something
– Internalize an understanding
• “The purpose of visualization is insight, not pictures”
– Insight: discovery, decision making, explanation
INFOVIS8803DV > SPRING 17
Bottom line
• Visuals help us think
– Provide a frame of reference, a temporary storage area
• Cognition Perception
• Pattern matching
• External cognition aid
Larkin & Simon ’87
Card, Mackinlay, Shneiderman ‘98
INFOVIS8803DV > SPRING 17
Great quote
“Contained within the data of any investigation is information
that can yield conclusions to questions not even originally
asked. That is, there can be surprises in the data…To regularly
miss surprises by failing to probe thoroughly with visualization
tools is terribly inefficient because the cost of intensive data
analysis is typically very small compared with the cost of data
collection.”
W. Cleveland
The Elements of Graphing Data
INFOVIS8803DV > SPRING 17
Part of our Culture
Seeing is
believingI see what
you’re saying
A picture is
worth a 1,000
words
INFOVIS8803DV > SPRING 17
The purpose of DataVis is ..
• Analysis – Understand your data better and act upon that
understanding
• Presentation – Communicate and inform others more effectively
INFOVIS8803DV > SPRING 17
When to Apply?
• Many other techniques for data analysis
– Statistics, DB, data mining, machine learning
• Visualization most useful in exploratory data analysis (EDA)
– Don’t know what you’re looking for
– Don’t have a priori questions
– Want to know what questions to ask
“A graphic display has many purposes but it achieves its highest value
when it forces us to see what we were not expecting.”
H. Wainer
INFOVIS8803DV > SPRING 17
EDA Examples
• Business
– Why has Hyundai made such great strides in the US market?
– How influential was their “Lose your job, we’ll buy the car back”
campaign?
– Have their cars improved in quality? If so, in what major ways?
– Is the Genesis as good of a car as the Lexus ES?
• Airlines
– What are the key factors causing flight delays in the US?
– Are delays worse in the summer or winter?
– Is the seasonal effect influenced by geographic location?
– How does competition at an airport affect flight delays?
INFOVIS8803DV > SPRING 17
More on EDA
“Information visualization is ideal for exploratory data
analysis. Our eyes are naturally drawn to trends,
patterns, and exceptions that would be difficult or
impossible to find using more traditional approaches,
such as tables or text, including pivot tables. When
exploring data, even the best statisticians often set their
calculations aside for a while and let their eyes take the
lead.”
S. Few
Now you see it
INFOVIS8803DV > SPRING 17
Tasks for DataVis?
• Search (OK)
– Finding a specific piece of information
• How many games did the Braves win in 1995?
• What novels did Ian Fleming author?
• Browsing (Better)
– Look over or inspect something in a more casual manner, seek
interesting information
• Learn about crystallography
• What has Jane been up to lately?
INFOVIS8803DV > SPRING 17
Key Benefits of Visualization
• Facilitating awareness and understanding
• Helping to raise new questions and supply answers
• Generating insights
• Telling a story and making a point
INFOVIS8803DV > SPRING 17
Key Challenge
• How to measure and prove?
– All those benefits are not easily quantifiable and measured
• Evaluation is perhaps primary open research challenge for
visualization*
* More on this in a later class
INFOVIS8803DV > SPRING 17
When to Apply?
• Visuals can frequently take the place of many words
• Visuals can summarize, aggregate, unite, explain, …
INFOVIS8803DV > SPRING 17
Data Visualization for BI is Big Business!
• 2003 Business Objects buys Crystal Decisions ($820M)• 2004 Agilent buys Silicon Genetics (GeneSpring, used in life sciences)• 2004 Hyperion buys QIQ Solutions, dashboard software) • 2005 General Dynamics buys MayaViz (logistics)• 2005 Business Objects acquires Infommersion (data visualization)• 2006 Actuate acquires Performancesoft (PBViews)• 2006 Microsoft buys Proclarity• 2007 Google buys Gapminder (demographics data Viz)• 2007 Cognos buys Celequest (business dashboards)• 2007 Business Objects buys Inxight• 2007 Oracle buys Hyperion $3.3B• 2007 SAP buys Business Objects ($6.8B)• 2007 IBM buys Cognos ($5B)• 2007 TIBCO buys Spotfire ($195M)• 2009 IBM buys SPSS• 2010 SAS acquires Memex (law enforcement, national security)• 2013 Tableau goes IPO ($254M)• 2015 Quid receives $53M in funding
(Some acquisitions involved more than DataVis products)
INFOVIS8803DV > SPRING 17
How do we decide on which type of
visual representation to use?
INFOVIS8803DV > SPRING 17
Tasks + Data
*People
INFOVIS8803DV > SPRING 17
INFOVIS8803DV > SPRING 17
Visualization Zoo (Heer et al.)
• Multivariate
• Geo-Based
• Time Series
• Hierarchies
• Networks
• Etc.
All visualizationsshare a common“DNA ”—a set ofmappings betweendata properties andvisual attributessuch as position,size, shape,and color—andcustomized speciesof visualizationmight always beconstructed byvarying theseencodings.
INFOVIS8803DV > SPRING 17
Multivariate Data: Scatterplot
INFOVIS8803DV > SPRING 17
Multivariate Data: Scatterplot Matrix
INFOVIS8803DV > SPRING 17
Multivariate Data: Parallel Coordinates
INFOVIS8803DV > SPRING 17
Time-Series Data: Index Chart
INFOVIS8803DV > SPRING 17
Time Series Data: Small Multiples
INFOVIS8803DV > SPRING 17
Time Series Data: Small Multiple (II)
INFOVIS8803DV > SPRING 17
Time-Series Data: Stacked Graph
INFOVIS8803DV > SPRING 17
Geo-Based Data: Maps
INFOVIS8803DV > SPRING 17
Hierarchies: Indented Tree
INFOVIS8803DV > SPRING 17
Hierarchies: Cartesian Node-Link
INFOVIS8803DV > SPRING 17
Hierarchies: TreeMap (Map of the Market)
Demohttp://www.marketwatch.com/tools/stockresearch/marketmap
INFOVIS8803DV > SPRING 17
Hierarchies: Sunburst
INFOVIS8803DV > SPRING 17
Networks: Force-Directed Layout
INFOVIS8803DV > SPRING 17
Networks: Arc Diagrams
INFOVIS8803DV > SPRING 17
Networks: Matrix Diagram
INFOVIS8803DV > SPRING 17
Types of Interactions*
• Details on Demand
• Focus + Context
• Dynamic Query
• Brushing & Linking
• Zoom & Pan
• Animation
• Change Representation (aka Re-encode)
* (not an exhaustive list but a good start :-) )
INFOVIS8803DV > SPRING 17
Overview & Detail – Mouse Selection
Clicking on an
item selects it
and attributes
of the data
point are shown
Selected item
Attributes
INFOVIS8803DV > SPRING 17
Overview & Detail – Pop-up Tooltips
Hovering mouse cursor brings up details of
item
INFOVIS8803DV > SPRING 17
Focus + Context (cont.)
INFOVIS8803DV > SPRING 17
Focus + Context
INFOVIS8803DV > SPRING 17
Dynamic Query
INFOVIS8803DV > SPRING 17
Brushing & Linking
• Highlighting connections in multiple views
– Simultaneously examine different attributes of a data
case
• In grid of scatter plots, select one “dot”, corresponding dots
highlighted in other scatter plots
– Simultaneously examine data case from different
views
• But need to keep straight where the data case is
INFOVIS8803DV > SPRING 17
Brushing & Linking
INFOVIS8803DV > SPRING 17
Brushing & Linking – Between Multiple Views
Sameitem
INFOVIS8803DV > SPRING 17
Brushing Example - DataMaps
Click on
histogram to
highlight states
UMD & VaTech
INFOVIS8803DV > SPRING 17
Zoom & Pan
• Plenty of Examples
– Google Maps, Google Earth
INFOVIS8803DV > SPRING 17
Change Representation
• Interactively change entire data presentation
– Looking for new perspective
– Limited real estate may force change
INFOVIS8803DV > SPRING 17
Change Representation – Example
Selecting different representation from options at bottom
INFOVIS8803DV > SPRING 17
Animation
• Time is actually time!
• Speed
• Timesteps
INFOVIS8803DV > SPRING 17
Recall the Gapminder Example
• What interaction
methods were used?
http://www.gapminder.org/world/
https://www.youtube.com/watch?v=jbkSRLYSojo
INFOVIS8803DV > SPRING 17
Interaction Methods in Gapminder
• Change Representation (Map, Chart)
• Linking/brushing/wiping
• Details-on-Demand (DoD)
• Animation
• Change Binding of Data to a Visual Representation
INFOVIS8803DV > SPRING 17
Homework 2 Due Next Class
• Find two visualizations of data from one of the suggested project domains. They can
be information presentations (static) or information visualizations (interactive). For
each visualization, do a one page write-up. Each page should have the visualization
(about half the page) and then:
1. What message(s) the visualization is intend to convey.
2. A critique of the visualization – list both pros and cons – ways in which the visualization does
a good job, ways in which it could be improved.
• I will ask some of you to show your visualization and share your pros and cons with
the class.
• Think of this assignment as a first step in identifying a potential project, which is why I
suggest that you be interested in the type of data being portrayed.
• Submit on T-Square AND please bring two (2) hard copies to submit in class.