1 18 April 2007 vizNET-LEEDS-PRES A Rough Guide to Data Visualization – Part 2 VizNET 2007 Annual Event Ken Brodlie School of Computing University of Leeds
2 18 April 2007 vizNET-LEEDS-PRES Outline of Session Visualizing Tables of Information Scatter plot matrices Parallel coordinates Glyphs Visualizing Structures of Information Interacting with a Visualization Focus and context Visualizing Time
3 18 April 2007 vizNET-LEEDS-PRES Visualizing Tables of Information Multivariate datasets
4 18 April 2007 vizNET-LEEDS-PRES Data Tables Multivariate datasets can be expressed as a data table Each entry in table is an observation An observation consists of values of a set of variables, or variates ABC variables observations
5 18 April 2007 vizNET-LEEDS-PRES What are these?
6 18 April 2007 vizNET-LEEDS-PRES Multivariate Visualization Example of iris data set 150 observations of 4 variables (length, width of petal and sepal) Check wikipedia for explanations of petals & sepals Techniques aim to display relationships between variables – the analytical task Challenge in visualization is to design the visualization to match the analytical task
7 18 April 2007 vizNET-LEEDS-PRES Scatter Plot For two variates, we have already met the scatter plot technique It is useful for showing what happens to one variable as another changes…
8 18 April 2007 vizNET-LEEDS-PRES Scatter Plot Matrix – Iris Data Set For a data table of N > 2 variables, we can look at all pairs of variables in a matrix of scatter plots. What can you observe?
9 18 April 2007 vizNET-LEEDS-PRES Parallel Coordinates – Iris Data One axis per variable Scaled to range Each observation is plotted as a ‘polyline’ Correlations may start to appear
10 18 April 2007 vizNET-LEEDS-PRES Parallel Coordinates Concept due to Alfred Inselberg Conceived the idea as a research student in 1959… … idea gradually refined over next 40 years
11 18 April 2007 vizNET-LEEDS-PRES The Screen Space Problem All techniques, sooner or later, run out of screen space Parallel co-ordinates Usable for up to 150 variates Unworkable greater than 250 variates Remote sensing: (5 variates, 16,384 observations)
12 18 April 2007 vizNET-LEEDS-PRES Brushing as a Solution Brushing selects a restricted range of one or more variables Selection then highlighted Note outlier
13 18 April 2007 vizNET-LEEDS-PRES Glyph Techniques – Star Plots Each observation represented as a ‘star’ Each spike represents a variable Length of spike indicates the value
14 18 April 2007 vizNET-LEEDS-PRES Glyph Techniques – Star Plots – Iris Data Each observation represented as a ‘star’ Each spike represents a variable Length of spike indicates the value Iris data set
15 18 April 2007 vizNET-LEEDS-PRES Multivariate Visualization Techniques Software: Xmdvtool Matthew Ward n Techniques designed for any number of variables –Scatter plot matrices –Parallel co-ordinates –Glyph techniques
16 18 April 2007 vizNET-LEEDS-PRES Visualizing Structures of Information
17 18 April 2007 vizNET-LEEDS-PRES Examples of Networks of Information My Window filestore Automobile web site - visualizing links
18 18 April 2007 vizNET-LEEDS-PRES Suppose user has three subdirectories: A, B and C First partition in X according to total size of each sub- directory ABC Tree Map of Filestore Note we are seeing the relative sizes of directories
19 18 April 2007 vizNET-LEEDS-PRES ABC Then within each subdirectory, we can partition in Y by the size of individual files, or further subdirectories Tree Map of Filestore
20 18 April 2007 vizNET-LEEDS-PRES Treemap Example Usenet news groups For history of treemaps see: hcil/treemap-history Developed over many years by Ben Schneiderman and colleagues
21 18 April 2007 vizNET-LEEDS-PRES Tree Maps – A Filesystem Download SequoiaView from: -Developed by Jarke van Wijk and colleagues Can be hard to decipher so..
22 18 April 2007 vizNET-LEEDS-PRES Visualizing Structure of Web Pages Star trees have been successfully used to visualize web site structures Place home page in centre with linked pages connected by hyperbolic arcs further arcs link to further links Escher woodcut….. inspired:
23 18 April 2007 vizNET-LEEDS-PRES Star Trees Automobiles web site Home page in centre Click on link you want...
24 18 April 2007 vizNET-LEEDS-PRES Star Trees Auto History moves to centre of screen Click on next link... And so on
25 18 April 2007 vizNET-LEEDS-PRES Star Trees Commercially available from Inxight Spun out of Xerox PARC
26 18 April 2007 vizNET-LEEDS-PRES Visualizing Web Searches
27 18 April 2007 vizNET-LEEDS-PRES Interacting with Visualizations
28 18 April 2007 vizNET-LEEDS-PRES Interaction Major difference between paper and computer-based visualization is ability to interact with picture and alter the presentation of the data We have already seen brushing and here we look at the idea of focus and context A recurring problem in visualization is lack of screen real estate Want to achieve: Focus: to see detail of immediate interest Context: to see the overall picture Want to do this interactively…
29 18 April 2007 vizNET-LEEDS-PRES Bifocal Display Magnify focus, demagnify context A possible application is in map reading This London underground application was developed by Marcelo Cohen in Leeds Zoom / pan Focus /context
30 18 April 2007 vizNET-LEEDS-PRES What is the Bifocal Display Doing? Transforming the information space to the display space Visual transfer functions Information space Display Space Normal display Information space Display Space Bifocal display context focus
31 18 April 2007 vizNET-LEEDS-PRES The Perspective Wall 2D layout wrapped around a 3D structure Space utilisation: -detail on centre panel 3x size of equivalent flat wall fitting field of view
32 18 April 2007 vizNET-LEEDS-PRES Multiple Views Multiple views can also be used to provide focus and context Powerpoint uses this to good effect!
33 18 April 2007 vizNET-LEEDS-PRES Time Visualizations
34 18 April 2007 vizNET-LEEDS-PRES Visual Metaphors : Long time periods Special techniques have been proposed for visualization over very long time periods Themeriver technique has been used to depict evolutionary behaviour… .. Possible with Excel Evolution of baby names.... Try it at: Laura and Martin Wattenberg
35 18 April 2007 vizNET-LEEDS-PRES Themeriver Themeriver for climate change… …
36 18 April 2007 vizNET-LEEDS-PRES River Metaphor Taglines Visualizing tags attached to Flickr online image sharing Evolution over time Show tags that are specific to a time period
37 18 April 2007 vizNET-LEEDS-PRES Cluster and Calendar based Visualization of Time Series Data Jarke van Wijk has shown how visualization can be used in analysis of time series data Opposite is power demand within ECN (Netherlands Energy Research Centre)… … hard to pick out patterns of usage
38 18 April 2007 vizNET-LEEDS-PRES Visualizing the Main Clusters By clustering the 365 days into 6 clusters of similar behaviour, we are able to visualize the key patterns of use… … but better still, in next slide we link to a calendar
39 18 April 2007 vizNET-LEEDS-PRES Calendar View of Power Demand
40 18 April 2007 vizNET-LEEDS-PRES Calendar View of Number of Employees at Work What can you observe? (NB Dec 5 th )
41 18 April 2007 vizNET-LEEDS-PRES Spiral Graphs Spiral graphs are a space- efficient way of visualizing long time series… From Alexa et al
42 18 April 2007 vizNET-LEEDS-PRES Time Wheel The Time Wheel allows several time series to be viewed simultaneously… … how successful is this? … rotation can help, why? … again cf parallel coordinates? Tominski, Abello, Schumann - Rostock