1	Exploring High Dimensional Data with the Rank-by-Feature Framework Ben Shneiderman ben@cs.umd.edu Founding Director (1983-2000), Human-Computer Interaction Lab Professor, Department of Computer Science Member, Institutes for Advanced Computer Studies & Systems Research University of Maryland College Park, MD 20742
2
3	Scientific Approach (beyond user friendly) Specify users and tasks Predict and measure time to learn speed of performance rate of human errors human retention over time Assess subjective satisfaction (Questionnaire for User Interface Satisfaction) Accommodate individual differences Consider social, organizational & cultural context
4	Design Issues Input devices & strategies Keyboards, pointing devices, voice Direct manipulation Menus, forms, commands Output devices & formats Screens, windows, color, sound Text, tables, graphics Instructions, messages, help Collaboration & communities Manuals, tutorials, training
5	U.S. Library of Congress Scholars, Journalists, Citizens Teachers, Students
6	Visible Human Explorer (NLM) Doctors Surgeons Researchers Students
7	NASA Environmental Data Scientists Farmers Land planners Students
8	Bureau of the Census Economists, Policy makers, Journalists Teachers, Students
9	NSF Digital Government Initiative Find what you need Understand what you Find
10	Information Visualization The eye… the window of the soul, is the principal means by which the central sense can most completely and abundantly appreciate the infinite works of nature. Leonardo da Vinci (1452 - 1519)
11	Using Vision to Think Visual bandwidth is enormous Human perceptual skills are remarkable Trend, cluster, gap, outlier... Color, size, shape, proximity... Human image storage is fast and vast Opportunities Spatial layouts & coordination Information visualization Scientific visualization & simulation Telepresence & augmented reality Virtual environments
12	Information Visualization: US Research Centers Xerox PARC 3-D cone trees, perspective wall, spiral calendar table lens, hyperbolic trees, document lens Univ. of Maryland dynamic queries, range sliders, starfields, treemaps, timeboxes, zoombars tight coupling, dynamic pruning, lifelines IBM, Microsoft, AT&T Georgia Tech, MIT Media Lab Univ. of Wisconsin, Minnesota, Calif-Berkeley, CMU Pacific Northwest National Labs
13
14
15
16
17
18
19
20	Information Visualization: Mantra Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand
21	Information Visualization: Data Types 1-D Linear Document Lens, SeeSoft, Info Mural, Value Bars 2-D Map GIS, ArcView, PageMaker, Medical imagery 3-D World CAD, Medical, Molecules, Architecture Multi-Var Parallel Coordinates, Spotfire, XGobi, Visage, Influence Explorer, TableLens, DEVise Temporal Perspective Wall, LifeLines, Lifestreams, Project Managers, DataSpiral Tree Cone/Cam/Hyperbolic, TreeBrowser, Treemap Network Netmap, netViz, SeeNet, Butterfly, Multi-trees
22	Treemap: view large trees with node values Space filling Space limited Color coding Size coding Requires learning
23	Treemap: Stock market, clustered by industry
24
25	Treemap: Gene Ontology
26
27
28
29
30	LifeLines: Customer Histories Temporal data visualization Medical patient histories Customer relationship management Legal case histories
31	Temporal Data: TimeSearcher 1.3 Time series Stocks Weather Genes User-specified patterns Rapid search
32	Temporal Data: TimeSearcher 2.0 Long Time series (>10,000 time points) Multiple variables Controlled precision in match (Linear, offset, noise, amplitude)
33	Goal: Find Features in Multi-D Data Finding correlations, clusters, outliers, gaps, à Cognitive difficulties in >3D Therefore utilize low-dimensional projections Perceptual efficiency in 1D and 2D Use Rank-by-Feature Framework to guide discovery
34	Multi-V: Hierarchical Clustering Explorer
35	What’s interesting?
36	What’s interesting?
37	Information Visualization: Tasks Overview Gain an overview of the entire collection Zoom Zoom in on items of interest Filter Filter out uninteresting items Details-on-demand Select an item or group and get details when needed Relate View relationships among items History Keep a history of actions to support undo, replay, and progressive refinement Extract Allow extraction of sub-collections and of the query parameters
38	Goal: Find Features in Multi-D Data Finding correlations, clusters, outliers, gaps, à Cognitive difficulties in >3D Therefore utilize low-dimensional projections Perceptual efficiency in 1D and 2D Use Rank-by-Feature Framework to guide discovery
39	Do you see anything interesting?
40	What features stand out?
41	Correlation…What else?
42	… and Outliers
43	Demonstration Breakfast Cereals 77 cereals 8 dimensions (or variables) : sugar, potassium, fiber, protein, etc. US counties census data 3138 counties 14 dimensions : population density, poverty level, unemployment, etc.
44	Rank-by-Feature Framework: 1D
45	Rank-by-Feature Framework: 2D
46
47	HCE Status In collaboration and sponsored by Eric Hoffman: Children’s National Medical Center Categorical Variables: 4.0 beta, May 2005 60K lines of C++ codes, 58 Classes 2,000+ downloads since April 2002 www.cs.umd.edu/hcil/hce
48	GRID Principles Graphics, Ranking & Interaction for Discovery (GRID) Study 1D, Study 2D, Then find features Ranking guides insight, Statistics confirm
49