Overview. interactive visualization of temporal data. Data types. Section A: introduction. 1-dimensional

Transcription

1 Overview interactive visualization of temporal data Wolfgang Aigner Version introduction what is special about the time dimension? what is temporal data? visualization roots excursus: art background taxonomy important concepts of time tasks for temporal data visualization classification infovis techniques presentation and discussion Vienna University of Technology interactive visualization of temporal data 00:2 Data types [Shneiderman, 1996] 1-dimensional Section A: introduction 2-dimensional 3-dimensional Temporal Multi-dimensional Tree Network = 4D space the world we are living in Vienna University of Technology interactive visualization of temporal data 00:4 1

2 Spatial + temporal dimensions Every data element we measure is related and often only meaningful when related to space + time Example: price of a computer where? when? Differences between space and time Space can be traversered arbitrarily we can move back to where we came from Time is unidirectional we can t go back or forward in time Humans have senses for perceiving space visually, touch Humans don t have senses for perceiving time Vienna University of Technology interactive visualization of temporal data 00:5 Vienna University of Technology interactive visualization of temporal data 00:6 Interactive visualization Gives us the ability to travel in time virtually. What do we consider as temporal data? General: Main focus on change over time of data elements More formally: [Schumann, 2003] Data elements are a function of time d = f(t) For discrete time steps: D = {(t1,d1), (t2, d2),, (tn, dn)} di = f(ti) Vienna University of Technology interactive visualization of temporal data 00:7 Vienna University of Technology interactive visualization of temporal data 00:8 2

3 Visualization roots Early time-series plot Statistics Visualization of time-series. The time-series plot is the most frequently used form of graphic design. [Tufte, 1983] Mostly one parameter over time. y t Art incorporating time more later Vienna University of Technology interactive visualization of temporal data 00:9 Part of a text for monastery schools 10th or 11th century (!) Inclinations of the planetary orbits over time 800 years before other time-series plots appeared Vienna University of Technology interactive visualization of temporal data 00:10 Train schedule Excursus: art background Paris to Lyon (1880s) Vienna University of Technology interactive visualization of temporal data 00:11 3

4 Renaissance [Masaccio and Masolino, Scenes from the Life of St. Peter, c , Brancacci Chapel, Florence] Multiple appearences of the same person within a single scene Vienna University of Technology interactive visualization of temporal data 00:13 Cubism The first documented occurrence of the fourth dimension being used in art appeared in 1910 in Paris. Origin: mathematics + physics (n-dimensional spaces) At this point, the fourth dimension was thought as time. Person walking down stairs --> Furth dimension in the painting by picturing different stages of the person s descent [Marcel Duchamp, Nude Descending a Staircase, 1912] Vienna University of Technology interactive visualization of temporal data 00:14 Cubism New ideas about the fourth dimension into the static domain of pictures. Overlays many different observations. Emphasizes process of looking and recording over time. [Picasso, Portrait of Vollard, 1910] Vienna University of Technology interactive visualization of temporal data 00:15 Comics Visual story telling over time. Many interesting techniques / paradigms. If you want to know more, start here: [Scott McCloud, Understanding Comics, 1994] Vienna University of Technology interactive visualization of temporal data 00:16 4

5 Reference to time [Schumann and Müller, 2000] Section B: taxonomy Time reference of data Example: temperature change of a lake is continuous over time vs. Time reference of presentation temperature measuring twice a year --> continuous change in real world --> discrete time points in presentation Vienna University of Technology interactive visualization of temporal data 00:18 Classification of visual representations Static representations Not time-dependent Does not automatically change over time Dynamic representations Time-dependent Changes dynamically over time Is a function of time Event-based representations [Schumann and Müller, 2003] Vienna University of Technology interactive visualization of temporal data 00:19 Interactivity Definitions: interactive information system An information system in which the user communicates with the computing facility through a terminal and receives rapid responses which can be used to prepare the next input. [McGraw-Hill Online Science Dictionary] interactive Of or relating to a program that responds to user activity. [American Heritage Online Dictionary] Of, relating to, or being a two-way electronic communication system (as a telephone, cable television, or a computer) that involves a user's orders (as for information or merchandise) or responses (as to a poll) [Merriam-Webster Online Dictionary] Interactive visualization!= Animation User controlled vs. data controlled Vienna University of Technology interactive visualization of temporal data 00:20 5

6 Tasks / Questions 1/2 [McEachren, 1995] Existence of a data element Does a data element exist at a specific time? Temporal location When does a data element exist on time? Is there any cyclic behavior? Time interval How long is the time span from beginning to end of the data element? Temporal texture How often does a data element occur? Tasks / Questions 2/2 Rate of change How fast is a data element changing or how much difference is there from data element to data element over time? Sequence In what order do data elements appear? Synchronization Do data elements exist together? [McEachren, 1995] Vienna University of Technology interactive visualization of temporal data 00:21 Vienna University of Technology interactive visualization of temporal data 00:22 Conceptual Organization Time series Time treated as linear sequence Don t confuse with linear time scale Time cycle [McEachren, 1995] Time treated as repeating cycle Many processes in nature and science have cyclic behavior e.g. days, years, seasons, Temporal dimensions Past Definite time - data element assignment Present Currently valid state Future Planning Temporal uncertainty Vienna University of Technology interactive visualization of temporal data 00:23 Vienna University of Technology interactive visualization of temporal data 00:24 6

7 GANTT charts 1/2 Section C: infovis techniques Contents: GANTT charts LifeLines Perspective Wall Calendar tools SpiraClock Temporal Objects Time Glyph Paint strips SOPOs ThemeRiver TM TimeWheel Lexis Pencil Serial Periodic Data Spiral Graph + Helix Intrusion Detection Time-wheel SW-Evolution Analysis Music Animation Machine Project management, project planning Tasks and their temporal attributes (location, duration) Milestones Past + present + future Hierarchical decomposition Vienna University of Technology interactive visualization of temporal data 00:26 GANTT charts 2/2 Pros: Well known representation Collapsable hierarchical decompostion Easy to comprehend Hundreds of tools available (i.e. MS Project) Cons: No uncertainty Space consumption (diagonal layout) LifeLines 1/2 [Plaisant et al., 1996, Plaisant et al., 1998] Based on Time Lines Facets Visualizing personal histories and patient information Horizontal bars showing temporal location and duration of data elements Past + Present Vienna University of Technology interactive visualization of temporal data 00:27 Vienna University of Technology interactive visualization of temporal data 00:28 7

8 LifeLines 2/2 Perspective Wall [Mackinlay et al., 1991] Pros: Simple and easy to comprehend Better layout than GANTT Use of vertical dimension Interactive time scale (zoom, pan) Cons: No hierarchical decomposition (only Facets) (Just past and present) Vienna University of Technology interactive visualization of temporal data 00:29 Large collections of documents Focus + Context of elements over time Intuitive 3D metaphor for distorting 2D layout Color coding Smooth transitions, 3D interactive animation Vienna University of Technology interactive visualization of temporal data 00:30 Calendar Tools SpiraClock 1/2 [Dragicevic and Huot, 2002] Past + present + future Calendar scale Events over time, repeating events Icons, Reminder Very well known (MS Outlook, ical, ) Interactive Techniques: Overview + Detail Zoom Filter Details on Demand Multiple Views Focus + Context Vienna University of Technology interactive visualization of temporal data 00:31 Visualization technique for nearby events. Intention: fill gap between static calendar and pop-up reminders. Continuous and non-intrusive feedback. Analog clock with white spiral inside representing near future. Vienna University of Technology interactive visualization of temporal data 00:32 8

9 SpiraClock 2/2 Interaction: Change time by moving hands. Adjust number of spiral revolutions (visibility of future events) Range: 1 hour - several days Not suited for all kinds of events i.e. conference, October Java applets and applications: Bus schedule, MS Outlook and vcal import Vienna University of Technology interactive visualization of temporal data 00:33 Temporal Objects 1/2 Depict future planning data with temporal uncertainty Starting instant (earliest start, latest start) Ending instant (earliest end, latest end) Maximum duration Minimum duration Based on LifeLines [Combi et al., 1999] Two encapsulated bars with caps at each end Vienna University of Technology interactive visualization of temporal data 00:34 Temporal Objects 2/2 Time Glyph 1/2 [Kosara and Miksch, 1999] Pros: Simple representation for complex time annotations Temporal uncertainty Easy to comprehend Cons: Only presentation, no interaction No direct manipulation Similar to Temporal Objects Additionally / Improvements: Time points are relative (Reference point) Notion for temporal granularity Notion for missing values / incomplete specifications Metaphor of bar lying on diamonds (preventing invalid constellations) User interaction / can be manipulated Vienna University of Technology interactive visualization of temporal data 00:35 Vienna University of Technology interactive visualization of temporal data 00:36 9

10 Time Glyph 2/2 Paint Strips [Chittaro and Combi, 2001] Metaphor of paint rollers Paint roller at the end of a line = line can expand Wall = expansion limit Smaller set of temporal attributes as Temporal Objects and Time Glyph Combination of strips (rope) Starting and finishing interval can t be defined independently from duration Vienna University of Technology interactive visualization of temporal data 00:37 Vienna University of Technology interactive visualization of temporal data 00:38 SOPOs 1/2 Rit s Set of Possible Occurences [Messner, 2000] 2D technique Area depicts set of valid (start, end) tuples Designed for easy graphical propagation of temporal constraints Cons: Representation more complicated than LifeLine based ones Space consumption SOPOs 2/2 Start interval: x-axis End interval: y-axis Minimum duration, maximum duration: constraining borders parallel to 45 time flow axis Vienna University of Technology interactive visualization of temporal data 00:39 Vienna University of Technology interactive visualization of temporal data 00:40 10

11 Intrusion Detection [Muniandy, 2001] Visualization of user access to machines over time. ThemeRiver TM 1/3 [Havre et al., 2000] Mapping: Time: circumference User: cylinder slice Machines: cubes on top Access: connection lines Annotations via tool tips (mouse hovering) Vienna University of Technology interactive visualization of temporal data 00:41 Visualize thematic variations over time. Across a large collection of documents. River Metaphor: the river flows through time. Changing width to depict changes. Themes or topics are colored currents. Vienna University of Technology interactive visualization of temporal data 00:42 ThemeRiver TM 2/3 Histogram vs. ThemeRiver TM : Discrete values Exact values Hard to follow a single current Continuous flow Interpolation, approximation Easy to follow a single current (curving continous lines) Vienna University of Technology interactive visualization of temporal data 00:43 ThemeRiver TM 3/3 User interaction: Hide or display topic + event labels time + event grid lines raw data points Choose alternate algorithms for line drawing Pan + Zoom Color relations Related themes are associated to the same color family Improvements: Parallel rivers Display of numeric values (on demand) Total number of documents Access documents directly User defined ordering Vienna University of Technology interactive visualization of temporal data 00:44 11

12 TimeWheel / Zeitrad 1/2 Time axis in the center Variable axis arranged circularly Lines connecting time and feature values Similar to parallel coordinates [Tominski et al., 2003] Variables parallel to time axis (upper and lower) can be explored most effectively Focus + Context by shortening of rotated axis and color fading TimeWheel / Zeitrad 2/2 User interaction: Rotation of variable axes (moving axes of interest into a position parallel to the time axis) Vienna University of Technology interactive visualization of temporal data 00:45 Vienna University of Technology interactive visualization of temporal data 00:46 MultiCombs [Müller and Schumann, 2003] Lexis Pencil [Francis and Pritchard, 1997] Axis based technique Multiple parameters on multiple time axis, circularly arranged Outward from the center of star-shaped Aggregated view of past values in the center Pencil-like geometric objects Mapping timedependent variables onto faces of the pencil Heterogenous data Can be located in 3D space to show the spatial context Tip allows exact positioning Problem: Occlusion Focus + Context On pencil: by radial arrangement In 3D space: enlarging pencil in focus Vienna University of Technology interactive visualization of temporal data 00:47 Vienna University of Technology interactive visualization of temporal data 00:48 12

13 Serial Periodic Data 1/6 [Carlis and Konstan, 1998] Serial Periodic Data 2/6 Visualize both, serial + periodic properties to reveal certain patterns Time continues serially, but weeks, month, and years are periods that reoccur Map time onto a spiral + spokes for orientation Data values are mapped to blots on spiral Area of blot proportional to value Pure serial periodic data Periods with constant durations Event-anchored serial periodic data Periods with different durations Start of a new period is indicated by an event Examples: Multi day racing data Project based time tracking Vienna University of Technology interactive visualization of temporal data 00:49 Vienna University of Technology interactive visualization of temporal data 00:50 Serial Periodic Data 3/6 Extension to 3D: Z-axis for different sets of data No quantitative meaning of z-axis Color coding of data sets Lidless, hollow cans Instead of blots Prevent occlusion Volume of can is proportional to data value Pro: good overview Cons: Occlusion Clutter Z-position meaningless Double mapping (z-pos + color) Vienna University of Technology interactive visualization of temporal data 00:51 Serial Periodic Data 4/6 User control: Rotation, zoom, pan, tilt Annotation features: Align different spirals vertically Definition of data derived border lines Display of several data sets simultaneously Using bar charts Color coded Multiple, linked spirals Vienna University of Technology interactive visualization of temporal data 00:52 13

14 Serial Periodic Data 5/6 Interval data Only duration of element Periodicity unknown Animation Serial Periodic Data 6/6 User experience findings: + Users quickly accept the notion of serial periodic data on a spiral + Users react to the spiral displays When they saw patterns, they tried to explain them by telling stories + Users want more Visualization sparked interest for further investigation - Tool not self explanatory Trained operator needed Vienna University of Technology interactive visualization of temporal data 00:53 Vienna University of Technology interactive visualization of temporal data 00:54 Spiral Graph 1/3 [Weber et al., 2001] Main intension: detection of periodic behavior Mapping data onto a spiral Mapping of data values to color and thickness of line Nominal + ordinal + quantitative data Spiral Graph 2/3 Two possibilities to detect periodic behavior: 1. Computational: Compute frequencies with higher amplitudes via Fourier Transformation 2. Visually: Utilize the visual system of a human observer to discover structures Spiral is animated by continously changing the cycle length Periodic behavior becomes immediately apparent (changing from unstructured to structured) User can stop animation when period is spotted 1 cycle = period length Vienna University of Technology interactive visualization of temporal data 00:55 Vienna University of Technology interactive visualization of temporal data 00:56 14

15 Spiral Graph 3/3 Extensions: Multi Spirals Compare a data set with cyclic patterns in other data. Rendering intertwined Spiral Graphs. 3D extension Problem: space mapping onto a helix. Brushing integrated. Selected region is displayed in 2D spiral. 3D helix best used for navigation only. Time-wheel 1/3 [Chuah and Eick, 1997] Visualization of software projects over time Multiple time-series placed in a circle Data attributes are color coded Global trends Helps to examine different trends within one object Easy recognition of two trends: Increasing trend Tapering trend Vienna University of Technology interactive visualization of temporal data 00:57 Vienna University of Technology interactive visualization of temporal data 00:58 Time-wheel 2/3 Time-wheel 3/3 Increasing trend Prickly fruit Tapering trend Hairy fruit Extension to 3D: Encodes the same attributes as the Time-wheel Uses height dimension to encode time Variables are encoded as slices of a base circle Pro: Easier to identify overall trends Cons: Occlusion Perspective Vienna University of Technology interactive visualization of temporal data 00:59 Vienna University of Technology interactive visualization of temporal data 00:60 15

16 Software Evolution Analysis [Jazayeri et al., 1999] Analyzing evolution of SWsystems / product families 3D visualization Colors encode versions Changes of parts over time Hierarchical decomposition Pattern analysis Not as information rich as Timewheel Vienna University of Technology interactive visualization of temporal data 00:61 Music Animation Machine (M.A.M.) 1/2 Online: Visualization of music Dynamic representation Relate audio to visual structure Simple representation for music extremely complex system Complex patterns [Malinowski] Vienna University of Technology interactive visualization of temporal data 00:62 Music Animation Machine (M.A.M.) 2/2 Each note is represented by a colored bar Each bar lights up as its note sounds The length of each bar corresponds exactly to the duration of its note as performed The vertical position of the bar corresponds to the pitch The horizontal position indicates the note's timing Vienna University of Technology interactive visualization of temporal data 00:63 Roundup Setting the scene Properties of time What are we talking about Tales about the past Early statistical graphics Time in art Looking backstage Ideas, concepts, definitions Opening the curtain State-of-the-art InfoVis techniques Vienna University of Technology interactive visualization of temporal data 00:64 16

17 Conclusions Temporal data covers a very broad field A lot of different techniques available Visualizations are task driven Cyclic/periodic behaviour is very common but relatively underexplored i.e. event-anchored data Not many dynamic techniques available Only very limited use of animation More interactivity is desireable Generally: Visualization sparks interest for further investigation Vienna University of Technology interactive visualization of temporal data 00:65 References 1/5 [Carlis and Konstan, 1998] Carlis, J.V. and Konstan, J.A., Interactive Visualization of Serial Periodic Data, ACM Symposium on User Interface Software and Technology, 1998 [Chuah and Eick, 1997] Chuah M.C. and Eick S.G., Glyphs for Software Visualization, International Workshop on Program Comprehension, pp , May [Combi et al., 1999] Combi, C., Portoni, L., and Pinciroli, F. (1999). Visualizing Temporal Clinical Data on the WWW. In Horn, W., Shahar, Y., and et al., editors, Proceedings of the Joint European Conference on Arti cial Intelligence in Medicine and Medical Decision Making (AIMDM'99), pages , Aalborg, Denmark.Springer, Berlin. [Dragicevic and Huot, 2002] Dragicevic, P. and Huot, S., SpiraClock: A Continuous and Non-Intrusive Display for Upcoming Events, CHI 2002, Interactive Poster: Visualization, 2002 Vienna University of Technology interactive visualization of temporal data 00:66 References 2/5 References 3/5 [Francis and Pritchard, 1997] Brian Francis and John Pritchard, Visualisationof historical events using Lexis pencils, Centre for Applied Statistics Fylde College, Lancester University, 1997 [Jazayeri et al., 1999] Jazayeri, M., Riva, C. and Gall, H., Visualizing Software Release Histories: The Use of Color and Third Dimension, Proceedings ICSM'99, Hongji Yang and Lee White (Ed.), IEEE Computer Society Press, [Kosara and Miksch, 1999] Kosara, R. and Miksch, S. (1999). Visualization Techniques for Time-Oriented, Skeletal Plans in Medical Therapy Planning. In Horn, W., Shahar, Y., Lindberg, G., Andreassen, S., and Wyatt, J., editors, Proceedings of the Joint European Conference on Arti cial Intelligence in Medicine and Medical Decision Making (AIMDM'99), pages , Aalborg, Denmark. Springer Verlag. Vienna University of Technology interactive visualization of temporal data 00:67 [Mackinlay et al., 1991] Mackinlay, J. D., Robertson, G. G., AND Card, S. K The Perspective Wall: Detail and context smoothly integrated, In Proceedings of CHI 91. ACM, New York, [Müller and Schumann, 2003] Müller, W. and Schumann, H., Visualization Methods for Time-Dependent Data - An Overview, Proceedings of the 2003 Winter Simulation Conference, S. Chick, P.J. Sanchez, D. Ferrin, and D.J. Morrice, eds., 2003 [Muniandy, 2001] Muniandy, K., Visualizing Time-Related Events for Intrusion Detection, Late Breaking Hot Topics Proceedings, InfoVis 2001 [Plaisant et al., 1996] Plaisant, C., Milash, B., Rose, A., Wido, S., and Shneiderman, B. (1996). LifeLines: Visualizing Personal Histories. In Proceedings CHI'96 ACM Conference on Human Factors in Computing Systems, pages , New York. ACM Press. Vienna University of Technology interactive visualization of temporal data 00:68 17

18 References 4/5 References 5/5 [Plaisant et al., 1998] Plaisant, C., Mushlin, R., Snyder, A., Li, J., Heller, D., and Shneiderman, B. (1998). LifeLines: Using Visualization to Enhance Navigation and Analysis of Patient Records. In Proceedings of the 1998 American Medical Informatic Association Annual Fall Symposium, pages [Schumann and Müller, 2000] Heidrun Schumann and Wolfgang Müller. Visualisierung - Grundlagen und allgemeine Methoden. Springer Verlag, Heidelberg, 2000 [Schumann and Müller, 2003] Schumann, H and Müller, W, Visualization Methods for Time-Dependent Data - An Overview, Proceedings of the 2003 Winter Simulation Conference, [Shneiderman, 1996] Shneiderman, Ben, The Eyes Have It: A Task by Data Type Taxonomy for Information Visualizations, Proceedings of the IEEE Symposium on Visual Languages, IEEE Computer Society Press, pp , Vienna University of Technology interactive visualization of temporal data 00:69 [Tominski et al., 2003] Tominski, Ch., Schulze-Wollgast, P. and Schumann, H., Visualisierung zeitlicher Verläufe auf geografischen Karten. Proc. GeoVis 2003, Hannover, 2003, 47-54, in German. [Tufte, 1983] Tufte, E.R., The Visual Display of Quantitative Informtion, Graphics Press, Cheshire, Connecticut, USA, [Weber et al., 2001] Weber, M., Alexa, M. and Müller, W., Visualizing Time-Series on Spirals, Proc. IEEE Symposium on Information Visualization 2001 (InfoVis 01), San Diego, USA, 7-13, 2001 Vienna University of Technology interactive visualization of temporal data 00:70 18