Visualizing Uncertainty Computer Science Perspective Ben Shneiderman Univ
Visualizing Uncertainty: Computer Science Perspective Ben Shneiderman, Univ of Maryland, College Park Alex Pang, Univ of California, Santa Cruz National Academy of Sciences Workshop March 3 -4, 2005, Washington, DC March 3 -4, 2005 NAS Workshop: Visualizing Uncertainty
March 3 -4, 2005 NAS Workshop: Visualizing Uncertainty
What do we mean by uncertainty? Why is this an issue now? March 3 -4, 2005 NAS Workshop: Visualizing Uncertainty
Sources of uncertainty Æ Measurement problems Æ Ranges/Summaries Æ Missing data Æ Human ratings Æ Potential deceptions Æ Privacy protection Æ Risk assessments Æ Forecasts 4 § Scientific data § Intelligence sources § Statistical analyses § Medical images § Gene expression § Simulations § Financial models § Weather § Consumer ratings NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Info. Viz Sci. Viz Visualizations l Text, statistical measures… l 1 D: Lists, documents, numeric ranges… 2 D: Geographic Info 3 D: Scientific Visualization l l l 5 Multi-Variate: Information Visualization Temporal: Patient histories, web logs … Tree: Taxonomies, org charts, directories … Network: Social, communication… NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Text: Statistical uncertainty l l l l 6 Text Poll Dems Reps Bush Kerry 1 D 2 D 3 D Multi-V Temporal Tree Network 14% 79% 81% 7% Margin of error +/- 3% Rain inches Frustration (N=372) Time Variables Time Lost (Incident) **. 293 Time to Fix (Incident) **. 233 Computer Years -. 041 Hours per Week *-. 124 * = p<. 05 ** = p<. 01 NAS Workshop: Visualizing Uncertainty Reliability SW 1. 0 low SE 1. 5 high NW 1. 2 low NE 0. 8 med March 3 -4, 2005
Risk/Danger vs. Trust/Validity/Security l Text l 1 D 2 D 3 D l l l 7 Multi-V Temporal Tree Network Terror Threat levels Risk Municipal Bonds low Blue Chip Stocks med Tech Stocks high Real Estate med Highly Unlikely NAS Workshop: Visualizing Uncertainty About Even Likely Highly Likely March 3 -4, 2005
1 D: Ranges, variations & forecasts l Text l 1 D 2 D 3 D l l l 8 Multi-V Temporal Tree Network NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Ball glyphs l Text l 1 D 2 D 3 D l l l Delta (observation - forecast) Multi-V Temporal Tree Network http: //www. cse. ucsc. edu/research/slvg/assim. html 9 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Arrow glyphs (Direction & velocity) http: //www. cse. ucsc. edu/research/avis/unvis. html 10 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Box glyphs Schmidt et al. , 2004, Underwater Environmental Uncertainty, IEEE CG&A http: //csdl. computer. org/comp/mags/cg/2004/05/g 5056 abs. htm 11 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Grid with transparency/shading Lefevre, Pfautz & Jones http: //ams. confex. com/ams/pdfpapers/82400. pdf 12 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Isolines with missing values Off-the-shelf software can give incorrect contours on data with lots of missing values. Modifications to contouring algorithm to account for large number of missing values. 13 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Pseudocolor shows mean values Luo, Kao & Pang, 2003, Euro. Vis http: //www. soe. ucsc. edu/~pang/op. pdf 14 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Darkness = uncertainty (high stddev) Mean = hue Skew = 1/saturation Stddev= 1/value 15 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Separate layer http: //www. cse. ohio-state. edu/~bordoloi/Pubs/pdf. Cluster. pdf 16 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Streamlines with binwise + Luo, Kao & Pang, 2003 http: //www. soe. ucsc. edu/~pang/op. pdf 17 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Weather forecast 18 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: NOAA Storm Prediction Center http: //www. spc. noaa. gov/products/ 19 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Dual views & grid lines Dark shows pollution Dark shows Certainty (Howard & Mac. Eachren, 1996) 20 NAS Workshop: Visualizing Uncertainty (Cedlink & Rhenigas, 2000) March 3 -4, 2005
Dual maps for Rate and reliability Bivariate color scheme Double hatch shows unreliable (Mac. Eachren et al. , 1998) 21 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
2 D: Gray for missing & interpolation Gray shows missing & interpolated value, superior to using black only Twiddy, R. , Cavallo, J. , and Shiri, S. 1994. Restorer: A visualization technique for handling missing data. IEEE Visualization 94, 212 -216 http: //svs. gsfc. nasa. gov/vis/a 000000/a 000010/index. html 22 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
3 D: Fuzzy molecular surface: HIV protease l Text l 1 D 2 D 3 D l l l Crisp molecular surface Probe radius 1. 4 Multi-V Temporal Tree Network Crisp molecular surface Probe radius 5. 0 23 Fuzzy molecular surface Probe radius 1. 4 Fuzzy molecular surface Probe radius 5. 0 Lee & Varshney (2002), UM Graphics and Visual Informatics Lab http: //www. cs. umd. edu/gvil NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
3 D: Fuzzy molecular densities 24 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
3 D: Uncertainty ‘dust’ 25 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
3 D: Color & opacity 26 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Multi-V: Database/spreadsheet tables l Text l 1 D 2 D 3 D l l l 27 Multi-V Temporal Tree Network NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Multi-V: Database/spreadsheet tables l Text l 1 D 2 D 3 D l l l 28 Multi-V Temporal Tree Network NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Temporal: Granularity of time l Text l 1 D 2 D 3 D l l l 29 Multi-V Temporal Tree Network NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Temporal: Granularity of time l Text l 1 D 2 D 3 D l l l 30 Multi-V Temporal Tree Network NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Temporal: Granularity of time l Text Time Uncertainty l l l l 31 1 D 2 D 3 D Multi-V Temporal Tree Network Hi Med Low NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Temporal: Granularity of time l Text l 1 D 2 D 3 D l l l 32 Multi-V Temporal Tree Network NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Tree: Topology, values & names l Text Certainty Hi Med l l l l 33 1 D 2 D 3 D Multi-V Temporal Tree Network Low Very Low - Gene Ontology - Tree of Life - Medical Subject Heading (Me. SH) - Chain of command/Org chart NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Tree: Topology, values & names l Text Certainty Hi Med l l l l 34 1 D 2 D 3 D Low Capacity Hi Med Multi-V Temporal Tree Network Low NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Tree: Topology, values & names l l l l 35 Text 1 D 2 D 3 D Multi-V Temporal Tree Network Leland or Lee Mike or Michael Alex or Alan Barbara Certainty Hi Med Low Scott Ben or Benjamin Diane or Di Ed or Edward or Eddie NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
March 3 -4, 2005 NAS Workshop: Visualizing Uncertainty
Network: Social relationship l Text l 1 D 2 D 3 D l l l Multi-V Temporal Tree Network http: //prefuse. sourceforge. net/demos-radial. html 37 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Network: Communication capacity 38 J. A. Brown, Mc. Gregor A. J and H-W Braun. NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Network: Node & edge uncertainty l Text Certainty Hi Med l l l 1 D 2 D 3 D Low Flow Hi Med Low l l Multi-V Temporal Tree Network Capacity Hi Med Low 39 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Next steps Explore novel approaches to: l Text: standard terms, percent, probabilities l Box plots, whiskers, ranges l Uncertainty glyphs, isoclines, … l Contours, surfaces, volume rendering l Hue, saturation, value, focus, haze, dust, … l Dual views, probes l Animation, blinking, shaking, flipping, … l Sound, haptics, … 40 NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
Next steps l l Heighten awareness of the problem among public, professionals, researchers & developers Support multi-valued data representation standards Explore techniques for each data type Develop guidelines for implementers: – – – l l l 41 Data formats Interactive interfaces Visual presentations Develop human performance evaluation methods Publish benchmark datasets & evaluation metrics Form guidelines for how to propagate/integrate uncertainty markers NAS Workshop: Visualizing Uncertainty March 3 -4, 2005
- Slides: 41