Required Readings Further Reading Big Picture Metric-Based Network Exploration and Multiscale Scatterplot. Hyperdimensional Data Analysis Using Parallel Coordinates. covered so far Yves Chiricota, Fabien Jourdan, Guy Melancon. Proc. InfoVis 04, Edward J. Wegman. Journal of the American Statistical design levels pages 135-142. Hierarchical Parallel Coordinates for Exploration of Association, Vol. 85, No. 411. (Sep., 1990), pp. 664-675. Lecture 11: Tabular Data problem, abstraction, encoding/interaction, algorithm methods Information Visualization Large Datasets Ying-Huey Fua, Matthew O. Ward, and Elke A. Parallel Coordinates: A Tool for Visualizing Multi-Dimensional taxonomy of visualization design concerns CPSC 533C, Fall 2011 Rundensteiner, IEEE Visualization ’99. Geometry. Alfred Inselberg and Bernard Dimsdale, IEEE Visualization ’90, 1990. Parallel sets: visual analysis of categorical data. Fabien Bendix, next stage: use these ideas for analysis and design Tamara Munzner Robert Kosara, and Helwig Hauser. Proc. InfoVis 2005, p 133-140. analyze previously proposed techniques and systems design new techniques and systems UBC Computer Science Mon, 17 October 2011 me: this lecture as example (and graphs/trees) you: project proposal, topic presentations 1 / 48 2 / 48 3 / 48 4 / 48 Analysis Via Levels and Methods Multiscale Scatterplots Problem and Abstraction Levels Abstraction Level: Data blur shows structure at multiple scales (problem characterization: generic network exploration) original data: relational network examples in this and graphs/trees lecture convolve with Gaussian minimal problem context; paper is technique-driven not links between Java classes note: only sometimes does this analysis occur in paper derived attributes: 2 structural metrics for network itself! slider to control scale parameter interactively problem-driven edge strength: cluster cohesiveness you need to interpret easily selectable regions in quantized image sw engr: logical dependencies between classes task abstraction: selection and filtering at different scales AppMetric vs Strength Scatterplot edges below color-coded by metric 5 within scatterplots (also something to do in your own project!) thus: table of numbers Strength Metric 4 3 2 1 0 1 2 3 4 Application Metric [Figs 3,4,5. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] [Fig 2. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] 5 / 48 6 / 48 7 / 48 8 / 48 Multiscale Scatterplot Selection Technique Multiscale Scatterplot Selection Technique Method: Linked Views Encoding/Interaction Level basic solution: new encoding: derived space created from original algorithm level: creating derived space second linked view: 3D node-link network visual encoding technique: scatterplots scatterplot image greyscale intensity is combination of patch selection in blurred scatterplot view shows mark: points. channels: horiz and vert position greyscale patches forming complex shapes corresponding components in network view blurred proximity relationships from original scatterplot interaction technique: range sliders to filter max/min enclosure of darker patches within lighter patches image: convolve with Gaussian filter selection in one view filters what is shown in the other limitations new interaction: point density in original scatterplot image interesting areas might not be easy to select as simple: sliders for filter size s and number of levels k quantize image into k levels rectangular regions, esp for complex derived attributes complex: single click to select all items > = k AppMetric vs Strength Scatterplot 5 Strength Metric 4 3 2 1 [Fig 3. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and 0 3 1 2 4 Application Metric Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] [Fig 3. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and [Fig 4. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and [Fig 6. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] 9 / 48 10 / 48 11 / 48 12 / 48 Results: IMDB Results: IMDB 2 Critique Critique strengths original data: IMDB graph single click in blurred scatterplot view selects entire clique successful construction and use of derived space metrics: network centrality, node degree appropriate validation 3 hubs selected in network view qualitative discussion of result images to show new technique capabilities synergy between encoding and interaction choices weaknesses somewhat tricky to follow thread of argument since intro/framing focuses on network exploration, but fundamental technique contribution more about scatterplot encoding/interaction [Fig 8. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and [Fig 7. Chiricota, Jourdan, and Melancon. Metric-Based Network Exploration and Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] Multiscale Scatterplot. Proc. InfoVis 2004, p 135-142.] 13 / 48 14 / 48 15 / 48 16 / 48
Recommend
More recommend
