Data Visualization

Angela Lu

Week 2
Information Systems, CB
City University of Hong Kong
Recap
• What is data visualization?
• Computer-based visualization systems provide visual
representations of datasets designed to help people carry out
tasks more effectively.
• Why visualize data?
• Intuitive way to show details; effective representation
• Is there a way to measure design effectiveness?
• Today: the what-why-how framework

2
 Analysis: What, why, and how
Analysis:• what
What,is shown? why, and how
– data abstraction
• why is the user looking at it?
• What is shown? – task abstraction
• data abstraction
• how is it shown?
• Why is the user looking at it?
– idiom: visual encoding and interaction
• task abstraction
• How is it shown?
abstract vocabulary
• Idioms: visual• encoding avoids domain-specific terms
and interaction
– translation process iterative, tricky
• what-why-how analysis framework as scaffold to think systematically
• What-why-how analysis framework
about design space as scaffold to think systematically
about design space
• Abstract vocabulary avoids domain-specific terms 11

• Translation process iterative, tricky

3
What is an idiom?
• An idiom is a distinct approach to creating and manipulating
visual representations.
• Examples: bar graph, pie chart, network etc…

4
 Analysis: What, why, and how
The big picture
• what is shown?
– data abstraction
• why is the user looking at it?
Data – task abstraction
physical data •type
how is it shown?
conceptual data– type
idiom: visual encoding and interaction
Mapping
Task • abstract vocabularyvisualavoids
encoding Image
domain-specific terms
questions, goals– translation process iterative, tricky visual channel, graphical
assumptions • what-why-how Processing marks to think systematically
algorithmsas scaffold
analysis framework
about design space
Idiom 11

metadata semantics
conventions

5
Why analyze?
• Imposes structure on huge design space
• Scaffold to help you think systematically about choices
• Analyzing existing as stepping stone to designing new
• Most possibilities ineffective for particular task/data combination

• So what design is effective?

• Example: Napoleon’s invasion of Russia

6
Minard's 1869 Napoleonic France's invasion of Russia 7
Single axis decomposition

8
 Y-axis: temperature (Q)
Mark decomposition

+ X-axis:Y-axis: temperature
longitude
+
over(O)
(Q) / time time

X-axis: longitude

=
=
Temp
Temp over space/time
over space / time (Q x Q)

9
+
Width: army size (Q)
X-axis: longitude
+
Y-axis: latitude
+
Width: army size

=
=

Army position and army size 10

 Y-axis: longitude (Q)

+ X-axis: latitude Mark

(Q) Composition

+
Width: army size (Q) Y-axis: temperature (Q)
+ X-axis: longitude (Q) / time (O)

= =
Temp over space/time (Q x Q)

Army position (Q x Q) and army size (Q)

Temp over space / time
Army position and army size
11
Depicts longitude, latitude, army size, and temperature. Any other features?
12
 Graphics (Proc. SIGGRAPH) 22:453– 462, 2003.]
Proc. InfoVis 2002, p 57–64.]

Analytical process
• What?
• Napoleon’s military movement in Russia 10

• Effect of Russia winter on military loss

• Why?
• Key variables related to loss?
• How?
• A few lines that represent key variables

13
 Value in cell

What?
What is shown: Data abstraction Geometry (Spatial)
Why?
Three major datatypes Position

How?
ThreeTypes
•Dataset datatypes: table, networks, spatial
[VAD Fig 2.1]
Tables Networks Networks
Spatial
Attributes (columns) Fields (Continuous) Geometry (Spatial)
Items Link Grid of positions
(rows)
Node
(item) Cell
Cell containing value Position
Node
(item)
Attributes (columns)
Multidimensional Table Trees
Value in cell

Value in cell • visualization vs computer graphics

–geometry is design decision 14
15
 Nodes, vertices

Edges, links, connections

16
17
 Value in cell

What?
Datatypes
Dataset and data types
Geometry (Spatial)
Why?
Position

Data and Dataset Types Spatial How?

Tables Networks & Fields Geometry Clusters, [VAD Fig 2.1]

Trees Sets, Lists
Items Items (nodes) Grids Items Items
Attributes Links Positions Positions
Attributes Attributes

Data Types
Items Attributes Links Positions Grids
Dataset Availability
Static Dynamic

18
17
 Value in cell

What?
Attribute
Attribute typestypes Geometry (Spatial)
Why?
Position

Attribute Types (Attributes are also called variables, dimensions) How?

Categorical Ordered [VAD Fig 2.1]
Ordinal Quantitative

Ordering Direction

Sequential Diverging Cyclic

19
 Ordinal Quantitative
Value in cell

What?
Attribute types Geometry (Spatial)
Why?
Position

How?
Ordering Direction
[VAD Fig 2.1]
Sequential Diverging Cyclic

16

20
21
Why visualize data?
• Visualization helps show details in the data
• Sometimes we don’t know exactly what questions to ask in
advance • {action, target} pairs
– discover distribution
• This is especially true in long-term exploratory analysis – compare trends

• Presentation of known results is made more fun – locate outliers

– browse topology
• Stepping stone towards automation

• External representation: perception vs cognition

• Intended task, measurable definitions of effectiveness

22
Actions: Search & query
• What does user know?
• Target, location • {action, target} pairs
• How much of the data matters? – discover distribution
– compare trends

• One, some, all – locate outliers

– browse topology

• Independent choices, mix & match

• Query, search, analyze

23
Higher-level actions: Analyze
Why?
• Why analyze? Actions Targets
• {action, target} pairs
Analyze All Data – discover distribution
– compare trends
Consume Trends Outliers
– locate outliers F
– browse topology
Discover Present Enjoy

Attributes
Produce
Annotate Record Derive One Many
tag Distribution Dependency

24
Derive
• Don’t just draw what you’re given!
• decide what the right thing to show is • {action, target} pairs
• create it with a series of transformations from the original dataset
– discover distribution
– compare trends
– locate outliers
– browse topology

• One of the four major strategies for handling complexity

• Example: trade balance

25
26
Derive

exports
imports
trade
balance

trade balance = exports −imports

Original Data Derived Data

27
What are the targets?
• Data & attributes
• How much data is enough? • {action, target} pairs
• What pattern? – discover distribution
– compare trends
• Network – locate outliers
– browse topology
• Topology
• Path
• Spatial
• Shape

28
Why?
Data Targets

All Data
Trends Outliers Features • {action, target} pairs
– discover distribution
– compare trends
– locate outliers
– browse topology

Attributes
Attributes: distribution, dependency, correlation, outliers…
One Many
Distribution Dependency Correlation Similarity

Extremes 29
How to identify outliers?

30
Features and anomaly detection
31
Target: Network Topology

• {action, target} pairs

– discover distribution
– compare trends
– locate outliers
– browse topology

32
Target: Spatial
Spatial data: shape or
distribution
• {action, target} pairs
– discover distribution
– compare trends
– locate outliers
– browse topology

33
ur levels of vis design
[A Nested Model of Visualization Design and Validation.
How to visualize? Munzner. IEEE TVCG 15(6):921-928, 2009
(Proc. InfoVis 2009). ]
ers?
• Idiom design: domain
cs of domain• How is it shown?of vis
to vocabulary abstraction
• Visual encoding: how to draw
ta abstraction
• Interaction: how to manipulate
king at it? task abstraction idiom

algorithm

om: how to draw

A Multi-Level Typology of Abstract Visualization Tasks
how to manipulate Brehmer and Munzner. 2013
[A Multi-Level Typology of Abstract Visualization Tasks
Brehmer and Munzner. IEEE TVCG 19(12):2376-2385,
2013 (Proc. InfoVis 2013). ] 34
 How?
Visualization:
Encode Encode Manipulate Facet

Arrange Change Juxtapose

Express Separate

Order Align Select Partition

Use
Navigate Superimpose

35
 from categorical and ordered
attributes
Use
Navigate
Color Superimpose
Visualization: Encode Hue Saturation Luminance

Map Size, Angle, Curvature, ...

from categorical and ordered
attributes
Color Shape
Hue Saturation Luminance

Motion
Size, Angle, Curvature, ... Direction, Rate, Frequency, ...

Shape

36
Analysis example: breakfast cereal

37
Analysis example: breakfast cereal

38
Analysis example: breakfast cereal

39
 Types
SpaceTree
What? Why? TreeJuxtaposer
? Why?
etworks & Fields Geometry Clusters, Ordered How?
How? [SpaceTree: Supporting Exploration in Large
Node Link Tree, Design Evolution and Empirical
ees Sets, Lists Ordinal Evaluation. Grosjean, Plaisant, and Bederson.
ee Tree Actions Actions SpaceTreeSpaceTree Proc. InfoVis 2002, p 57–64.]

Analysis: Example
ms (nodes) Grids Items Items
nks Present
PresentPositions
Positions Locate Locate
Identify Encode Encode
Identify Navigate Navigate
Select
[TreeJuxtaposer: Scalable Select
Filter
Tree Comparison Using Focus Filter
AggregateAggregate
+Context With Guaranteed Visibility. ACM Trans. on
tributes Attributes Quantitative Graphics (Proc. SIGGRAPH) 22:453– 462, 2003.]

Why?
What? Why? Actions How? Targets
Targets Targets TreeJuxtaposer
TreeJuxtaposer
Ordering Direction
Tree Path between
Networks Actions
Fields
Path (Continuous)
Analyze between
two Encode SpaceTree
nodestwo nodesSequential All Data
Encode
Navigate Navigate
Select Select
Arrange Arrange
mns) Present
Consume
Grid of positions
Locate Identify Encode
Trends Navigate
Outliers FeaturesSelect Filter Aggregate
Discover Present Enjoy
Link
Cell
What? Node Why? Diverging
How?
(item)
value Attributes (columns) Attributes
Tree Actions
Produce SpaceTree
Targets TreeJuxtaposer
One Many
Present
Annotate
Value in cell
Locate
Record Identify
Derive
Cyclic Encode Navigate Select Filter Aggregate
l Table Trees
Path between two nodes
tag
Encode
Distribution
Navigate
Dependency
Select
Correlation
Arrange
Similarity

Extremes
Search
lue in cell
Targets TreeJuxtaposer
Target known Target unknown
Path between two nodes Encode Navigate Select Arrange
Location
known
Lookup Browse Network Data
Location Locate Explore Topology 24
al) unknown

Query Paths

n Identify Compare Summarize

What? 40
 Types
SpaceTree
What? Why? TreeJuxtaposer Express Separate
? Why?
etworks & Fields Geometry Clusters, Ordered How?
How? [SpaceTree: Supporting Exploration in Large
Node Link Tree, Design Evolution and Empirical
ees Sets, Lists Ordinal Evaluation. Grosjean, Plaisant, and Bederson.
ee Tree Actions Actions SpaceTreeSpaceTree Proc. InfoVis 2002, p 57–64.]

Analysis: Example Select

ms (nodes) Grids Items Items
Order Align
nks Present
PresentPositions
Positions Locate Locate
Identify Encode Encode
Identify Navigate Navigate
Select
[TreeJuxtaposer: Scalable Select
Filter
Tree Comparison Using Focus Filter
AggregateAggregate
+Context With Guaranteed Visibility. ACM Trans. on
tributes Attributes Quantitative Graphics (Proc. SIGGRAPH) 22:453– 462, 2003.]

Why?
What? Why? Actions How? UseTargets
Navigate
Targets Targets TreeJuxtaposer
TreeJuxtaposer
Ordering Direction
Tree Path between
Networks Actions
Fields
Path (Continuous)
Analyze between
two Encode SpaceTree
nodestwo nodesSequential All Data
Encode
Navigate Navigate
Select Select
Arrange Arrange
mns) Present
Consume
Grid of positions
Locate Identify Encode
Trends Navigate
Outliers FeaturesSelect Filter Aggregate
Link
Cell
Discover Present Enjoy Map
What? Node Why? Diverging
How? from categorical and ordered
value
(item)
Attributes (columns) Attributes attributes
Tree Actions
Produce SpaceTree
Targets TreeJuxtaposer
One Color
Many
Present
Annotate
Value in cell
Locate
Record Identify
Derive
Cyclic Encode Navigate Select Filter Aggregate
l Table Trees Hue Saturation Luminance
Path between two nodes
tag
Encode
Distribution
Navigate
Dependency
Select
Correlation
Arrange
Similarity

Extremes
Search Size, Angle, Curvature, ...
lue in cell
Targets TreeJuxtaposer
Target known Target unknown
Path between two nodes Encode Navigate Select Arrange
Location
known
Lookup Browse Network Data
Location Topology Shape
Locate Explore 24
al) unknown

Query Paths Motion

n Identify Compare Summarize Direction, Rate, Frequency, ...
What? 41
Next…
• Tableau demo

42
What is Tableau?
• Tableau: one of the fastest growing business analytics and data
visualization tools; easy to share analytics in the cloud
• Good for fast analytics
• Big data; built-in calculations
• Live data update and connection
• Little programming required; drag and drop visualization

• A short demo …
• https://www.youtube.com/watch?v=YfE9jBq002s&ab_channel=T
ableauSoftware
43
44
A Tableau example
• Let’s check out some of the Tableau features….
• Tableau Public Gallery example: popular artist in UK
• https://public.tableau.com/en-gb/gallery/artist-most-no-1-
albums-uk?tab=viz-of-the-day&type=viz-of-the-day
• Example workbook uploaded on Canvas

45
Data & file structures of Tableau
• Tableau data types include dimensions and measures
• Tableau uses a Workbook and sheet file structure.
• A Workbook contains sheets, which can be a Worksheet, a
Dashboard, or a Story.

46
Tableau Data Types: Dimensions &
Measures
• Dimensions: fields that cannot be aggregated; usually used
for row or column headings
• Measures: fields that can be calculated, aggregated, or used
for mathematical operations; usually used for plotting or
giving values
• Dimensions & measures can be converted into each other
(drag, right-click)

47
Tableau Worksheet

• A Worksheet contains a single

view along with shelves, legends,
and the Data pane.
• From worksheet we can access
Source Data Dimensions,
Measures, Custom Fields.

48
Tableau Dashboard
• A Dashboard is a collection of views from multiple
worksheets.
• From Dashboard we can access Worksheet but we can not
access dimensions & measures directly.

49
Tableau Story
• A Story contains a
sequence of worksheets or
dashboards that work
together to convey
information.
• From Story we can access
Dashboard and
Worksheets but we can not
access dimensions
measures directly.

51