![Page 1: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/1.jpg)
Lecture 7: Statistical GraphicsInformation VisualizationCPSC 533C, Fall 2007
Tamara Munzner
UBC Computer Science
1 October 2007
![Page 2: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/2.jpg)
Readings Covered
Visual information seeking: Tight coupling of dynamic query filterswith starfield displays. Chris Ahlberg and Ben Shneiderman, ProcSIGCHI ’94, pages 313-317
Metric-Based Network Exploration and Multiscale Scatterplot. YvesChiricota, Fabien Jourdan, Guy Melancon. Proc. InfoVis 04, pages135-142.
Graph-Theoretic Scagnostics. Leland Wilkinson, Anushka Anand,and Robert Grossman. Proc InfoVis 05
The Visual Design and Control of Trellis Display. R. A. Becker, W. S.Cleveland, and M. J. Shyu Journal of Computational and StatisticalGraphics, 5:123-155. (1996).
Multi-Scale Banking to 45 Degrees. Jeffrey Heer, Maneesh Agrawala.IEEE TVCG 12(5) (Proc. InfoVis 2006), Sep/Oct 2006, pages701-708.
![Page 3: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/3.jpg)
Statistical Graphics
I long history for paper-based views of dataI springboard for infovis
I interacting with scatterplotsI interactive dynamic queriesI multiscale structureI matrix of scatterplots, level of indirectionI linked views (more on this next time)
I ordering dot plotsI improving line charts
![Page 4: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/4.jpg)
Scatterplots
I encode two input variables with spatial position
I show positive/negative/no correllation between variables
[http://upload.wikimedia.org/wikipedia/commons/0/0f/Oldfaithful3.png]
![Page 5: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/5.jpg)
Dynamic Queries on ScatterplotsI tight coupling: immediate feedback after actionI starfield = interactive scatterplotI dynamic queries as lightweight visual exploration
I vs. composing SQL query
[Visual information seeking: Tight coupling of dynamic query filters with starfielddisplays. Chris Ahlberg and Ben Shneiderman, Proc SIGCHI ’94, p 313-317][http://www.cs.umd.edu/hcil/pubs/screenshots/FilmFinder/]
![Page 6: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/6.jpg)
FilmFinder
[Visual information seeking: Tight coupling of dynamic query filters with starfielddisplays. Chris Ahlberg and Ben Shneiderman, Proc SIGCHI ’94, p 313-317][http://www.cs.umd.edu/hcil/pubs/screenshots/FilmFinder/]
![Page 7: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/7.jpg)
FilmFinder
[Visual information seeking: Tight coupling of dynamic query filters with starfielddisplays. Chris Ahlberg and Ben Shneiderman, Proc SIGCHI ’94, p 313-317][http://www.cs.umd.edu/hcil/pubs/screenshots/FilmFinder/]
![Page 8: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/8.jpg)
FilmFinder
[Visual information seeking: Tight coupling of dynamic query filters with starfielddisplays. Chris Ahlberg and Ben Shneiderman, Proc SIGCHI ’94, p 313-317][http://www.cs.umd.edu/hcil/pubs/screenshots/FilmFinder/]
![Page 9: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/9.jpg)
FilmFinder
[Visual information seeking: Tight coupling of dynamic query filters with starfielddisplays. Chris Ahlberg and Ben Shneiderman, Proc SIGCHI ’94, p 313-317][http://www.cs.umd.edu/hcil/pubs/screenshots/FilmFinder/]
![Page 10: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/10.jpg)
Critique
I clear successesI fast, lightweight visual queriesI details on demandI easy to use for novices
I more arguable: alphaslidersI other techniques: data vis sliders, fisheye menus,
speed-dependent automatic zooming
![Page 11: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/11.jpg)
Multiscale Scatterplots
I blur shows structure at multiple scalesI convolve with GaussianI slider to control scale parameter interactively
I easily selectable regions in quantized image
[Metric-Based Network Exploration and Multiscale Scatterplot. Yves Chiricota, FabienJourdan, Guy Melancon. Proc. InfoVis 04]
![Page 12: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/12.jpg)
Metric-Based Exploration: Software EngrI linked views for metric-based exploration
I graph viewI axis 1: strength metric (topological graph structure)I axis 2: software engr metric (public methods)
[Metric-Based Network Exploration and Multiscale Scatterplot. Chiricota, Jourdan, andMelancon. Proc. InfoVis 04]
![Page 13: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/13.jpg)
Metric-Based Exploration: IMDB
I axis 1: centrality, for locating cliquesI axis 2: node degree, for size of cliqueI axis 2: clustering index
[Metric-Based Network Exploration and Multiscale Scatterplot. Chiricota, Jourdan, andMelancon. Proc. InfoVis 04]
![Page 14: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/14.jpg)
Critique
I interesting followup to Wattenberg paperI exploiting perceptual mechanisms
I suitable for intermediate/expert analysisI abstraction might be difficult for novice use
![Page 15: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/15.jpg)
SPLOM: Scatterplot Matrix
I show all pairwise variable combos side by side
I matrix size grows quadratically with variable count
[Graph-Theoretic Scagnostics. Wilkinson, Anand, and Grossman. Proc InfoVis 05.]
![Page 16: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/16.jpg)
Graph-Theoretic ScagnosticsI reduce problem to constant size
I overview matrix of 9 geometric metricsI meta-SPLOM: each point represents scatterplot
I detail on demand to see individual scatterplots
Graph-Theoretic Scagnostics. Leland Wilkinson, Anushka Anand, and RobertGrossman. Proc InfoVis 05.
![Page 17: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/17.jpg)
Measuring Scatterplots
I aspects and measuresI outliers: outlyingI shape: convex, skinny, stringy, straight
I computed with convex hull, alpha hull, min span tree
I trend: monotonicI density: skewed, clumpyI coherence: striated
[Graph-Theoretic Scagnostics. Wilkinson, Anand, and Grossman. Proc InfoVis 05.]
![Page 18: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/18.jpg)
Measuring Scatterplots
[Graph-Theoretic Scagnostics. Wilkinson, Anand, and Grossman. Proc InfoVis 05.]
![Page 19: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/19.jpg)
Results
[Graph-Theoretic Scagnostics. Wilkinson, Anand, and Grossman. Proc InfoVis 05.]
![Page 20: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/20.jpg)
Results
[Graph-Theoretic Scagnostics. Wilkinson, Anand, and Grossman. Proc InfoVis 05.]
![Page 21: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/21.jpg)
Critique
I very powerful and elegant methodI curse of dimensionality is hard problem
I abstraction level clearly appropriate for experts
![Page 22: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/22.jpg)
Automatic Dotplot Ordering: Trellisalphabetical site,variety use group median
[The Visual Design and Control of Trellis Display. Becker, Cleveland, and Shyu. JCSG5:123-155 1996]
![Page 23: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/23.jpg)
Trellis Structure
I conditioning/trellising: choose structureI pick how to subdivide into panelsI pick x/y axes for indiv panelsI explore space with different choices
I multiple conditioningI ordering
I large-scale: between panelsI small-scale: within panels
I main-effects: sort by group medianI derived space, from categorical to ordered
![Page 24: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/24.jpg)
Confirming Hypothesis
I dataset error with Morris switched?
I old trellis: yield against variety givenyear/site
I new trellis: yield against site andyear given variety
I exploration suggested by previousmain-effects ordering
[The Visual Design and Control of Trellis Display. Becker,Cleveland, and Shyu. JCSG 5:123-155 1996]
![Page 25: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/25.jpg)
Partial Residuals
I fixed dataset, Morris data switchedI explicitly show differences
I take means into accountI line is 10% trimmed mean (toss
outliers)
[The Visual Design and Control of Trellis Display. Becker,Cleveland, and Shyu. JCSG 5:123-155 1996]
![Page 26: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/26.jpg)
Critique
I careful attention to statistics and perceptionI finding signals in noisy data
I trends, outliers
I exploratory data analysis (EDA)I Tukey work fundamental, Cleveland continues
![Page 27: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/27.jpg)
Banking to 45 Degrees
I mentioned but not explainedin Trellis paper
I previous work by ClevelandI perceptual principle: most
accurate angle judgement at45 degrees
I pick aspect ratio(height/width) accordingly
[www.research.att.com/∼rab/trellis/sunspot.html]
![Page 28: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/28.jpg)
Multiscale Banking to 45I frequency domain analysisI find interesting regions at multiple scales
[Multi-Scale Banking to 45 Degrees. Heer and Agrawala, Proc InfoVis 2006vis.berkeley.edu/papers/banking]
![Page 29: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/29.jpg)
Choosing Aspect Ratios
I FFT the data, smooth byconvolve with Gaussian
I find interestingspikes/ranges in powerspectrum
I cull nearby regions if toosimilar, ensure overviewshown
I create trend curves foreach aspect ratio
![Page 30: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/30.jpg)
Multiscale Banking to 45
[Multi-Scale Banking to 45 Degrees. Heer and Agrawala, Proc InfoVis 2006vis.berkeley.edu/papers/banking]
![Page 31: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/31.jpg)
Presentation Topics Due Oct 19
I pick three topics that you wantI optional: veto one of the three days
I Nov 5 or Nov 7 or Nov 19I send me email by Oct 19
I Subject: 533 submit topics
![Page 32: Lecture 7: Statistical Graphicstmm/courses/cpsc533c-07... · Lecture 7: Statistical Graphics Information Visualization CPSC 533C, Fall 2007 Tamara Munzner UBC Computer Science 1 October](https://reader035.vdocuments.net/reader035/viewer/2022063011/5fc4ef1ccc284054872897f6/html5/thumbnails/32.jpg)
TopicsI application domains
I software vizI computer networks vizI db/datamine vizI cartographic vizI social networks vizI ...
I data domains
I time-series data vizI text/document collection
vizI tree/hierarchy
visualizationI graph drawingI high dimensional dataI ...
I techniques/approaches
I interactionI focus+contextI navigation/zoomingI glyphsI animationI brushing/linkingI statistical graphicsI ...
I other
I frameworks/taxonomiesI perceptionI evaluationI ....