Module 4 - Visualization Idioms and Interaction

Question 1

Visualisation Idioms (V1)

Answer 1

visualisation of tabular data

idioms are described by:
1. data
no. categorical attributes
no. qualitative attributes
2. mark
3. channel
4. tasks
- Arrange
express,separate,order,align,use
- Manipulate
change,select,navigate

Question 2

Idioms with n keys and 1 value (V1)

Answer 2

• Bar Chart (1 Key)
  Mark = lines
  Channel = length, spatial region
  Tasks = Compare, lookup values
  Scalability = hundreds
• Stacked Bar Chart (2 Key)
  Mark = stack of line marks
  Channel = length, spatial region (first bar is aligned, others are not), color
  Tasks = Compare, lookup values, part-to-whole relationship
  Scalability = 1 to 12 for stacked attribute
• Pie Chart (1 Key)
  Mark = Separated area color
  Channel = Angle for quantitative, Color for categorical
  Tasks = Part to whole judgement
  Scalability = 1 to 12

Question 3

Line Chart (V1)

Answer 3

Data = 1 Ordered Attribute (Key), 1 Quantitative Attribute (Value)
Mark = Points, line connecting points
Channels = Aligned lengths to express quantitative value, Separated and ordered by key attribute into horizontal regions
Tasks = Trend, show relationship (Connection marks emphasize ordering of items along key axis)

Question 4

Why choose (Stacked) Bar chart vs Line Chart? (V1)

Answer 4

Depends on key attribute, if categorical or ORDERED. otherwise violates expressiveness principles

Question 5

Streamgraph (V1)

Answer 5

Data = 1 Categorical attribute (KEY), 1 Quantitative Attribute, 1 ORDERED attribute (KEY)
Marks = height/layers which represents count
Channel = color
Tasks = Find trends, Part to whole relationship
Scalability = Better then stacked bar charts

Question 6

Heatmap (V1)

Answer 6

Data = 2 Categorical Keys, 1 Quantitative Attribute,
Mark = Separate and align in 2d matrix, indexed by the 2 keys
Channel = Color by quantitative attribute
Tasks = Find clusters, outliers, patterns

Question 7

Statistical Value Idioms - One numerical Attribute (V2)

Answer 7

1. Histogram
   - size/no. of bins!
   Mark = Line
   Channels = Length encodes frequency
   Tasks = Distribution
2. Boxplot
   - median,min,max,Q1 and Q3
   Mark = Lines/box
   Channels = Length is derived values
   Tasks = Distribution
   * does not show frequency, which is dangerous
3. Violinplot
   - median,min,max,Q1,Q3 AND density
   Mark = Lines/box
   Channels = Length is derived values, width is frequency
   Tasks = Distribution

Question 8

Information Visualization (V3)

Answer 8

The use of computer-supported, interactive, visual representations of abstract data to amplify cognition

Question 9

Why interaction? (V3)

Answer 9

Pros
- Too much data to show in one view
- Different audiences with different questions
- Increases engagement
Cons
- Takes time to learn
- Takes time to use
- Getting lost

Question 10

Interaction Principles (V3)

Answer 10

Low latency feedback
- if longer then 10 sec, show progress bar
Overview,Zoom/Filter,Details on Demand

Question 11

Interaction Techniques (V3)

Answer 11

1. Change Over Time
   - consider using animation to show changes
2. Select
   - click vs hovering
   - box vs lasso selection
3. Manipulate
   a. Item Reduction
   - Zoom
   Geometric (camera) and Semantic (zoom level)
   - Pan/Translate
   - Constrained
   b. Attribute Reduction
   - Slice
   show items that match specific values by attribute
   - Cut
   show items within a range
   - Project
   dimensionality reduction

Question 12

Multiple views (V4)

Answer 12

multiple encodings showing the same data with different perspectives, while taking into account:
design choices (view count, visibility, arrangement)
linking&brushing (two way interaction)

Question 13

Brushing Vs Filtering (V4)

Answer 13

Brushing = Show all missing and selected
Filtering = Show selected (with no context)

Question 14

Types of Multiple Views (V4)

Answer 14

1. Multiple views
   - different visual encodings, same data.
2. Small Multiples
   - same encoding, different data
3. Overview & Detail
   - Same visual encoding, same data, different zoom-level
   - Detail on demand: extra view with more info on selection
4. Focus + Context
   Different levels of detail integrated in the same view
   - Show area/items of interest (focus) in detail and surroundings (context) in less detail
   - Distortion techniques