1 / 25

Data Visualization

Data Visualization. Visualization: The use of computer-supported, interactive, visual representations of data to amplify cognition. Information Visualization: The use of computer-supported, interactive visual representations of abstract data to amplify cognition. S. Card .

bachyen
Télécharger la présentation

Data Visualization

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Data Visualization Visualization: The use of computer-supported, interactive, visual representations of data to amplify cognition. Information Visualization: The use of computer-supported, interactive visual representations of abstract data to amplify cognition. S. Card

  2. Data Visualization Brief History Key Techniques Science versus Aesthetics

  3. Data Visualization: Brief History • Jacques Bertin • Semiology of Graphics: Diagrams, Networks, Maps, 1983 • coined the term “using vision to think” Literature Overview:

  4. Data Visualization: Brief History • Edward Tufte • The Visual Display of Quantitative Information, 2001 • Envisioning Information, 1990 • Visual Explanations: Images and Quantities, Evidence and Narrative, 1997 • The Cognitive Style of PowerPoint, 2006 • Keywords: visualization of statistical data, cartograms, history of information visualization, visualization displays, micro and macro readings, small multiples, escaping flatland • Tufte Home Page • Tufte Article on Stanford Alumni Magazine Literature Overview:

  5. Data Visualization: Brief History maximization of useful information on a limited display Literature Overview:

  6. Data Visualization: Brief History • Chaomei Chen • Information Visualization: Beyond the Horizon, 2004 • Mapping Scientific Frontiers: The Quest for Knowledge Visualization, 2003 • Information Visualisation and Virtual Environments, 1999 • CiteSpace II: Detecting and visualizing emerging trends and transient patterns in scientific literature, 2006 • Top 10 unsolved information visualization problems, 2005 (pdf) • Keywords: domain visualization, network/graph visualizations, visualization in virtual (collaborative) environments, social networks • Chaomei Chen Home Page Literature Overview:

  7. Data Visualization: Brief History • Ben Shneiderman • The Craft of Information Visualization: Readings and Reflections, 2003 • Readings in Information Visualization: Using Vision to Think, 1999 • Keywords: human factors, HCIL, visual dynamic query tools, social networks • Film Finder • Ben Shneiderman on Social Networks Literature Overview:

  8. Data Visualization: Brief History • Stuart Card • A Framework for Visualization, 2002 • The Internet Edge: Social, Technical, and Legal Challenges for a Networked World, 2000 • Readings in Information Visualization: Using Vision to Think, 1999 • The Structure of the Information Visualization Design Space (survey paper on evaluation) • Keywords: HCI, Model Human Processor, GOMS (goals, operators, methods, and selection rules) theory of user interaction, information foraging theory, statistical descriptions of Internet use • Stuart Card Bio • GOMS Literature Overview:

  9. Data Visualization: Brief History More on HCI: Affordances Hick’s Law Fitts’ Law Five Hat Racks Usability Engineering Evaluation Literature Overview:

  10. Data Visualization: Scientific • GIS, Geographic Data Visualizations • Therese-Marie Rhyne • Daniel Keim • Alan MacEachren • Waldo Tobler (cartograms survey paper) • Andre Skupin (cartograms & perception) Literature Overview:

  11. Data Visualization: Scientific • Network / Graph Visualization • Peter Eades • Thomas Fruchterman, Edward Reingold • Tomihisa Kamada, Satoru Kawai • Graph Drawing: Algorithms for the Visualization of Graphs • Stephen Eick • Kenneth Cox • Richard Becker (Visualizing Network Data) • Tamara Munzner (H3viewer) • John Lamping, Ramana Rao (Focus+Context) • George Furnas (Fisheye View) • Graph Drawing Survey Paper Literature Overview:

  12. Data Visualization: Aesthetics • Martin Wattenberg • Director of Visual Communication Lab at IBM Watson Center Data Visualization & Aesthetics:

  13. Data Visualization: Aesthetics • Ben Fry • Organic information design (Anemone) • Software visualization • (Dismap) • Keywords: Qualitative versus quantitative representation of data, • algorithmic design, processing, genetic algorithms Data Visualization & Aesthetics: Haplotypes

  14. Data Visualization: Aesthetics • Golan Levin • Golan Levin Home • Lisa Jevbratt • Lisa Jevbratt Projects Data Visualization & Aesthetics:

  15. Data Visualization: Methods • Methods and Algorithms • MDS, • SOM, • Force-Directed Placement, • Grand tour, • Parallel Planes, • Glyphs, • Node and link displays, • Tree maps, • Matrix representations, • Cone trees, • Fisheye views, • Focus+context views, • Cartograms… Data Visualization

  16. Data Visualization: Algorithms • Proposed to achieve several aesthetic criteria about graph layouts • Uniform distribution of nodes • Uniform edge lengths • Minimum edge crossings • Symmetry • demo Force-Directed Placement

  17. Data Visualization: Algorithms • Peter Eades proposed as a heuristic approach. • The idea is to calculate attractive forces between connected nodes and repulsive forces between every pair of nodes. • Force models varied significantly: • Eades: was complex to run in real time • Fruchterman, Reingold: reduced complexity of Eades’ equations • Kamada Kawai: based on Hooks’ law and minimization of energy • Iterative algorithms Force-Directed Placement

  18. Data Visualization: Algorithms • Ms thesis, Basak Alper Force-Directed Placement

  19. Data Visualization: Algorithms • Brief intro • A method for dimensionality reduction, enables to visualize 40 dimensional data on a 2D display • The idea is to keep distance relations between nodes, proportionally consistent as you reduce dimensions of the space • If distance in 40D space is d, then distance in 2D space should be λd , where λis a constant for all elements MDS

  20. Data Visualization: Algorithms • Multi-dimensional scaling • Metric MDS methods based on eigen value analysis of the matrix showing relatedness of every element • Non-iterative and very costly • If distance in 40D space is d, then distance in 2D space should be λd , where λis a constant for all elements MDS

  21. Data Visualization: Algorithms • Non-metric MDS is proposed by Kruskal to overcome problems with metric MDS • Non-metric MDS defines a stress function to place data nodes on lower dimensional space • Nodes are displaced to lower stress and iterations are stopped when overall stress reaches below a certain threshold • Stress function MDS where d ij is the distance in high dimensional space and g ij is the distance in low dimensional space distance function is generally the Euclidian distance

  22. Data Visualization: Algorithms • Kohonen self-organizing maps • Another way of reducing dimensions of data in a neural networks fashion • Pseudocode for the algorithm: • 1. Initialize Map: Randomize the map's nodes' weight vectors • 2. Grab an input vector • 3. Traverse each node in the map • 1. Use Euclidean distance to find similarity between the input vector and the map's node's weight vector • 2. Track the node that produces the smallest distance • 4. Update the nodes in the neighbourhood by pulling them closer to the input vector (neighborhood function) • 5. Increment t and repeat while t < λ (total number of iterations) SOM

  23. Data Visualization: Algorithms • Initialize map: Create a matrix of vectors, where the size of these vectors is equal to the dimensions of data and magnitudes are in the range of data • The initial map can be totally random or organized in a certain way, for instance magnitudes • Neighborhood function: is generally a Gaussian, and the radius is generally reduced over iterations SOM

  24. Data Visualization: Algorithms • Resulting arrangement of the vectors on the map is based on similarities of input data vectors. • For each vector in higher dimensions, the position of the closest vector on the map is its position in lower dimensional space (which is generally 2D) SOM

  25. Data Visualization: Algorithms • Processing demo • Model Tunes SOM

More Related