INFORMATION VISUALIZATION CMPT 481 How to represent data? How effective is a visual attribute in conveying information? 1 2 3 4 E.g. How might number be best conveyed? Information Visualization What is visualization? Information visualization basics Visual variables “The power of the unaided mind is highly overrated. Without external aids, memory, thought and reasoning are all constrained. But human intelligence is highly flexible and adaptive, superb at inventing procedures and objects that overcome its own limits. The real power comes from devising external aids that enhance cognitive abilities.” Donald Norman What is visualization? Understanding and seeing “To visualize” Previously: “to construct a visual image in the mind” Now: “a graphical representation of data or concepts” Definition of visualization: The use of computer-supported visual representations of data to amplify cognition “Using vision to think” The purpose of visualization is insight, not pictures Anscombe’s quartet Anscombe’s quartet Anscombe’s quartet What InfoViz is not Not scientific visualization where data is not abstract History of Visualization William Playfair (1780’s) First systematic charting of abstract data line graph bar chart pie chart circle graph Napoleon's march on Moscow Example http://www.smartmoney.com/map-of-the-market/ Example http://www.babynamewizard.com/voyager Important: Perceptual Foundations We can take advantage of pre-attentive processing: But should avoid illusions: Distortions Ambiguities Paradoxes Hallucinations Mapping Data to Visual Form Types of attribute data Nominal data Ordinal data Ordering and ranking based on < and > e.g. restaurant ratings Interval data Category data that can only be compared for equality e.g. apple, orange Ordering and arithmetic possible but no natural zero e.g. dates, temperature Ratio data Ordered, natural zero e.g. height, weight, age, length Selecting the Mappings Most variables are mapped on to only 6 high-level visual encodings Spatial (most important) Marks Connection Enclosure Retinal Properties Temporal Properties Adding interactivity View transformations to increase information content Location probes use location to reveal additional info e.g. tool tips, detail screen Viewpoint controls zoom and pan Mapping data to graphics Data table determine what is to be visualized filter data appropriately Determine visual structure choose mappings for attributes important horizontal and vertical don’t overmap visual variables most determine appropriate scales for axes Determine interactivity probes and viewpoint controls Where to live in the U.S. City Climate Housing HlthCare Crime Transp Educ Abilene 521 6200 237 923 4031 2757 Akron 575 8138 1656 886 4883 Albany 468 7339 618 970 Albany 476 7908 1431 Albuquerque 659 8393 Alexandria 520 Allentown Arts Recreat Econ Long Lat Pop 996 1405 7633 -99.689 32.559 110932 2438 5564 2632 4350 -81.518 41.085 660328 2531 2560 237 859 5250 -84.158 31.575 112402 610 6883 3399 4655 1617 5864 -73.7983 42.7327 835880 1853 1483 6558 3026 4496 2612 5727 -106.65 35.083 419700 5819 640 727 2444 2972 334 1018 5254 -92.453 31.302 135282 559 8288 621 514 2881 3144 2333 1117 5097 -75.4405 40.6155 635481 Alton 537 6487 965 706 4975 2945 1487 1280 5795 -90.1615 38.794 268229 Altoona 561 6191 432 399 4246 2778 256 1210 4230 -78.395 40.515 136621 Amarillo 609 6546 669 1073 4902 2852 1235 1109 6241 -101.849 35.383 173699 Ab ile ne Ak ro n Al ba ny Al ba Al ny bu qu er qu Al e ex an dr ia Al le nt ow n Al to n Al to on a Am ar illo Climate Climate 700 600 500 400 300 200 100 0 Climate Crime vs. climate 700 600 Climate 500 400 300 200 City 100 Linear (City) 0 0 200 400 600 800 Crim e 1000 1200 1400 1600 Multiple attributes – radar plot HousingCost 8000 7000 Econ Climate 6000 5000 4000 3000 2000 Recreat HlthCare 1000 0 Arts Crime Amarillo Albany Educ Transp Geographical orientation Spatial Where something occurs in space Is perceptually dominant Good for discriminating values and finding patterns Mappings: Unstructured (no axis, just something present or not) Nominal (a region is divided into subregions, and can be present or not) Ordinal (the ordering of subregions is also meaningful) Quantitative (the regions has a metric) Spatial Examples Marks Marks are visual things that occur in space Points (0D) Lines (1D) Areas (2D) Volume2 (3D) Connection and Enclosure Allow relationships to be shown many other objects Bubble Sets http://www.youtube.com/watch?v=P6CgBmIiXaE Retinal Properties Encode other dimensions (attributes) of the data Temporal Maps some data to time E.g. animation of the state of water as temperature goes from +125 to -10 degrees Celsius Temperature is mapped to time Often, time is mapped to time (compressed) http://www.nytimes.com/interactive/2009/09/12/bu siness/financial-markets-graphic.html But do our mappings work? visualizations have a specific purpose Mappings must be expressive Encodes only intended data relations and no other data relations Certain visual encodings lend themselves to more uses • E.g., size can be used for nominal, ordinal, or quantitative • E.g., shape can only be used for nominal data Caveats Lie Factor Chart Junk Lie Factor The representation of numbers, as physically measured on the surface of the graphic itself, should be directly proportional to the quantities represented. [Tufte, 1991] Calculating Lie Factor Chart Junk The interior decoration of graphics generates a lot of ink which does not tell the viewer anything new. The purpose of the decoration varies - to make the graphic appear more scientific, to enliven the display, to give the designer an opportunity to exercise artistic skill. Regardless of the cause, it is all non-data-ink or redundant data-ink, and it is often chartjunk." [Tufte, 1983] Chart Junk Data-Ink Ratio ‘Data ink’: the essential non-erasable ink used to present the data ‘Non-data ink’: the redundant ink used to elaborate or decorate the graph The Data-Ink Ratio is defined as the percentage: (100 x Data-ink) / (Total ink used on graphic) Low data-ink ratio High Data-Ink Ratio So what is wrong with this? Is this better? Some research has shown that both types of charts can be equally read without errors and take about the same amount of time to read the junky charts are remembered significantly better over a long period of time including the existence of the chart and chart details junky charts to provide a clearly biased message If you are interested… Bateman, S., Mandryk, R.L., Gutwin, C., Genest, A.M., McDine, D., Brooks, C. 2010. Useful Junk? The Effects of Visual Embellishment on Comprehension and Memorability of Charts. In ACM Conference on Human Factors in Computing Systems (CHI 2010), Best paper award. http://hci.usask.ca/publications/view.php?id=173 Viz Techniques Relationship Among Data Points ScatterPlot Network Diagram Matrix Chart Viz Techniques: Small multiples Learn the representation once, use it many times Invites comparisons Viz Techniques Compare a set of values Bar Chart Block Histogram Bubble Chart Viz Techniques Track Rise and Falls Over Time Line Graph Stack Graph w\ categories Viz Techniques Parts of a Whole Pie Chart Tree Map For comparison Viz Techniques Analyze Text Tag Cloud Word Tree PhraseNet Viz Techniques Geographic Data Map Overlays Vis Techniques Hierarchical Data Tree Maps Radial Layouts Vis Techniques N-Dimensional Parallel Coordinates Other Interaction Techniques Linking and Brushing A new Take on Linking and Brushing VisLinks: Linking across different visualizations More Info and Links Interesting visualizations http://www.informationisbeautiful.net/ General Viz Software and Toolkits Prefuse (http://prefuse.org/) Ggobi (http://www.ggobi.org/) ManyEyes (http://manyeyes.alphaworks.ibm.com/) InfoViz News and Resource: Information aesthetics (http://infosthetics.com/) InfoVis Wiki (http://www.infovis-wiki.net/)
© Copyright 2026 Paperzz