COMP 150-04
Visualization
Lecture 11
Assignment 5: Maps
Design a thematic map visualization
Option 1: Choropleth Map
Implementation in Processing
Option 2: Tourist Map
Design/interaction sketch + thorough discussion Option 3: Interactive Layered Map
or “Mapper’s Delight”
Implementation in Processing
Visual information-seeking mantra
“Overview first, zoom and filter, then details on demand.” Design of GUIs and interactions
Ben Schneiderman, “The eyes have it: A task by data type taxonomy for information visualization”
Desktop interfaces
Interactions we take for granted can be powerful Detail on demand:
Mouse selection
Tangible interfaces
Novel interaction styles Detail on demand:
Gestural selection Multiple selections
Interaction in infovis
Static or dynamic visualization? What are the goals?
What aspects of the design can we control? What user tasks/operations must we support?
Static infovis
Goal: Create an effective, expressive view of the data Data encoding
Composition
Perception: popout, Gestalt Cognitive skills
Communicate Compare, rank
Static infovis
Goal: Create an effective, expressive view of the data Data encoding
Composition
Perception: popout, Gestalt Cognitive skills
Communicate Compare, rank
Identify correlation, causation
Goal: Enable user to focus on goals rather than controls
Presentation: Good static views linked together well
Perception
Cognitive skills Motor skills Explore
Find best match
Obtain the data
Order the data into categories by meaning
Remove all but the data of interest
Discern patterns, place the data in mathematical context
Select a visual encoding model
Improve the basic representation
Support dynamic queries
ACQUIRE PARSE FILTER MINE REPRESENT REFINE INTERACT PARSE FILTER MINE PRESE REFIN TERAC
ACQUIRE PARSE FILTER MINE REPRESENT REFINE INTERACT PARSE FILTER MINE PRESE REFIN TERAC DATA HANDLING GRAPHIC DESIGN INTERACTION DESIGN Regular expressions, ... Perl, Python, ... Graphics APIs UI toolkits Visualization toolkits
What is “interactive”?
< 10 sec cognitive response
< 1 sec system response, conversation break < 0.1 sec visual continuity, GUI widgets
Data type taxonomy
1D, 2D, 3D Temporal Multi-dimensional (nD) Tree NetworkBen Schneiderman, “The eyes have it: A task by data type taxonomy for information visualization”
Task taxonomy
Overview: see overall patterns in data Zoom: see a subset of data
Filter: see a subset based on values Detail on demand: see values of items Relate: compare values
History: keep track of actions Extract: mark and capture
Ben Schneiderman, “The eyes have it: A task by data type taxonomy for information visualization”
Task taxonomy
Ben Schneiderman, “The eyes have it: A task by data type taxonomy for information visualization”
Visual Languages, 1996
Overview: see overall patterns in data Zoom: see a subset of data
Filter: see a subset based on values Detail on demand: see values of items Relate: compare values
History: keep track of actions Extract: mark and capture
overview+detail focus+context geometric zoom semantic zoom mouseover query selection query brushing/linking dynamic query
Overview+Detail display
Overview+Detail display
Overview+Detail display
Overview+Detail display
Show overview and detail in separate views
+ No spatial distortion
Focus+Context display
Unified view:
Focus object is in full detail Surrounding, contextual info
is available with less detail
+ Simultaneous display matches human visual system
- Distortion/occlusion may impede understanding
Patrick Baudisch, “Focus plus context screens” http://patrickbaudisch.com
Pan and zoom
Geometric vs. semantic zoom? Distortion?
Semantic zoom
Hybrid views: drill down to display more information
+ Simultaneous display of overview and detail possible - Visual clutter: occlusion may impede understanding
Ken Perlin, Zoomable user interfaces http://mrl.nyu.edu/~perlin/experiments/zoom/Presentation.html
Recall: Small multiples
Pictorial and tabular layouts Constancy of design
Same design structure repeated for all images
Economy of perception
Draws the eye to differences and outliers
Recall: Small multiples
Invite comparison, contrasts
Must use same units, scale, measurements
Coordinated multiple views
Use two or more views to support understanding of one concept Vary views by visual encoding, scale, data set
Different visual encodings of the same data
Different scale of same data, same encoding (overview+detail) Different data with same encoding, same scale (small multiples)
Coordinated multiple views
TimeSearcher: Visual Exploration of Time Series Data http://www.cs.umd.edu/hcil/timesearcher
Brushing
TimeSearcher: Visual Exploration of Time Series Data http://www.cs.umd.edu/hcil/timesearcher
Linking
TimeSearcher: Visual Exploration of Time Series Data http://www.cs.umd.edu/hcil/timesearcher
Coordinated multiple views
Addresses issue of scale: can’t fit many marks/attributes in one view Addresses issues of data complexity
Design considerations:
Attention: Working memory, context switch Learnability
Screen real estate
Operations on data tables
Rearrange by attribute Sort by attribute
Select a subset of records
Write a query: formal query language SELECT address
FROM bostondb
WHERE price <= 500,000 AND bedrooms >= 2
bathrooms >= 2 AND garage == true
Dynamic queries
Visual model of the world: Objects
Actions: rapid, incremental, reversible Query: Direct selection
Results: “Immediate” (< 0.1 sec)
Ben Shneiderman et al, Dynamic HomeFinder, U. Maryland, 1993 http://www.youtube.com/watch?v=5X8XY9430fM
Dynamic queries on the web
http://housingmaps.com http://housingmaps
Dynamic queries on the web
Dynamic queries on the web
Dynamic queries
+ Responsive interaction: “fly through the data” + Natural interaction: find the “best” results + Exploration
- Conjunctive controls: requires user training - Spatially expensive
Designing and evaluating a program
for molecular visualization
Dynamic queries: replace query language Multiple views: show multiple alignment
Variation: data types, encodings, resolution Conciseness
Linking and brushing Attention management