Your SlideShare is downloading. ×
0
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Iscram ebert keynote_apr_2012_v2_small
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Iscram ebert keynote_apr_2012_v2_small

772

Published on

Integrative Visual Analytics for Effective Decision Making and Action (Much More Than …

Integrative Visual Analytics for Effective Decision Making and Action (Much More Than
Big Data Analytics with Visualization Added)
David S. Ebert
Keynote Presentation
Tuesday, April 23, 9:00 am-10:00 am
Asia Pacific Hall

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
772
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
3
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. Visual Analytics: Powering Discovery, Innovation, and Decision Making (Much more than Big Data Analytics plus Visualization) David S. Ebert ebertd@purdue.edu
  • 2. PathIntroduction to our Center - VACCINEMotivation and introduction to visual analyticsExample solutions we’ve developed Health visual analytics Maritime public safety and law enforcement Public safety Resource allocationVisual analytic challenges and future April 2012
  • 3. SFU, JIBC UBC UW Purdue Penn St. Ind U VaTechStanford NC UNCC A&T Navajo GaTech Tech JSU U Stuttgart UT UHD Swansea U Austin FIU April 2012
  • 4. VACCINE Mission• Provide visual analytic and scalable solutions for all 2.3 million extended homeland security personnel • 185,000 DHS personnel, 350,000 law enforcement personnel, 750,000 homeland security practitioners, 2.3 million extended personnel• Achieve excellence in visual analytics and visualization sciences• Educate homeland security stakeholders and the next generation of talent April 2012
  • 5. Who We Are:International Team of Experts- 75+ Faculty, 25 institutions • Purdue University • University of Houston, Downtown • Georgia Institute of Technology • Virginia Tech • Pennsylvania State University • Indiana University • Stanford University • Florida International University • University of North Carolina at • University of Texas at Austin Charlotte • Morgan State University • University of Washington • Navajo Technical College • Simon Fraser University • University of Stuttgart • University of British Columbia • University of Swansea, UK • Justice Institute of British • Oxford Columbia • University of Calgary • Ontario Institute of Technology • University of Manitoba • Dalhousie University • Carleton University • April 2012 University of Victoria
  • 6. Some VACCINE Regional andCorporate Partners• Indiana Department of Homeland Security • Motorola• Indiana State Department of Health • Harris Corporation• Indianapolis Public Safety • Boeing• Charlotte Mecklenburg Police Department • Next Wave Systems, LLC• Lafayette, West Lafayette, Purdue Police • Banfield, The Pet Hospital Departments • Raytheon• Tippecanoe County EMA & Sherriff’s • MacDonald, Dettwiler and Associates Department • Oculus Info Inc.• Harrisburg, PA Police • Kx Systems• Coast Guard Sector Boston, Seattle, LA • Bank of America• Joint Harbor Operations Command • Duke Energy Center • World Vision International• Port of Seattle • Gates Foundation• Indianapolis Public Safety • Kimberly Clark• USCG District 9, D5, D1 • Hallmark April 2012
  • 7. Some VACCINE Government Partners• U.S. Coast Guard • National Maritime Intelligence• Federal Emergency Management Center Agency • Defence Research & Development Canada• Customs and Border Patrol• National Geospatial Intelligence • Foreign Broadcast Information Agency Service • US-CERT• National Science Foundation• Army Research Office • Oak Ridge National Laboratories• Department of Defense • US Army Corps of Engineers- ERDC• Department of Health and Human Services • Pacific Northwest National Laboratory• U.S. Dept. of State • ARL CERDEC & ARDEC• National Institute of Justice • US Attorney’s Office (David Capp) April 2012
  • 8. What We Do• Enable effective decision making through interactive visual analytic environments• Enable effective communication of information• Provide quantitative, reliable, reproducible evidence• Enable user to be more effective from planning to detection to response to recovery• Enable proactive and predictive visual analytics• Enable effective situational awareness (perception, comprehension, projection) April 2012
  • 9. PathIntroduction to our Center - VACCINEMotivation and introduction to visualanalyticsExample solutions we’ve developed Health visual analytics Maritime public safety and law enforcement Public safety Resource allocationVisual analytic challenges and future April 2012
  • 10. MotivationTo solve today’s and tomorrow’s problems requires exploring, analyzing, and reasoning with massive, multisource, multiscale, heterogeneous, streaming data Image of Analyst’s Notebook April 2012
  • 11. DefinitionVisual Analytics1 is the science of analytical reasoning facilitated by interactive visual interfacesPeople use visual analytics tools and techniques to • Synthesize information and derive insight from massive, dynamic, ambiguous, and often conflicting data • Detect the expected and discover the unexpected • Provide timely, defensible, understandable assessments • Communicate assessment effectively for action 1. Illuminating the Path: The R&D Agenda for Visual Analytics, Editors: Thomas and Cook April 2012
  • 12. Statistical and Information  Analytics Interaction Science Geospatial Analytics Cognitive &  Scientific Computing Perceptual Sciences  Foundations of  Visual Visual Analytics Graphics and Visualization  Design and  Communication  Decision Sciences  Knowledge Knowledge Discovery  Data Management &  Knowledge April 2012
  • 13. What’s Needed for Proactive andPredictive Visual Analytics?• Reliable and reproducible models and simulation• Understanding of the data • Distribution and skewness, errors, appropriate analysis techniques• Understanding of the sources and types of data• Comparable or Correlative sources of data • Appropriate transformations applied to enable meaningful comparison and correlation• Understanding of the use and problem to be solved! April 2012
  • 14. How to Harness Visual Representations• Visual representations translate data into a visible form, highlighting important features, such as commonalities and anomalies• Visual representations make it easy for users to perceive salient aspects of their data quickly• These visual representations augment the cognitive reasoning process with perceptual reasoning which enhances the analytical reasoning process 1 – J. J. Thomas and K. A. Cook (Ed.). Illuminating the Path: The R&D Agenda for Visual Analytics. National Visualization and Analytics Center, 2005. April 2012
  • 15. Why Was I Invited?Visual Analytics Usesfor Crisis Management and Response?• Trend analysis, clustering, anomaly detection• Interactive, multi-day, month, type investigation• Multisource, multimedia data integration & analysis• Purpose: • Planning for resiliency • Long-term analysis • Predictive analytics • Training • Detection • Investigation • Response • Recovery, remediation April 2012
  • 16. Visual Analytics vs. Situational AwarenessSituational Awareness Sensemakingsituational awareness as a sensemaking process (courtesy Alan MacEachren) Situational awareness model (after Endsley, 1995; top) compared to model of sense making (after Pirolli & Card, 2005; bottom) Situational Awareness Perception of Comprehension State of Projection of Performanc elements in of current Decision environment future status e of actions current situation situation goals & objectives; preconceptions Sensemaking Foraging loop Sense-making loop Search & filter Read & extract Schematize Build case Tell storyExternal data Evidence Shoebox Schema Hypotheses Presentation source file Search for Search for Search for Search for support Reevaluate information relations evidence April 2012
  • 17. PathIntroduction to our Center - VACCINEMotivation and introduction to visual analyticsExample solutions we’ve developed Health visual analytics Maritime public safety and law enforcement Public safety Resource allocation and business intelligenceVisual analytic challenges and future April 2012
  • 18. Example: Public Health Visual Analytics Health Pandemic ZoonosesSurveillance Simulation and Planning Planning and Response Map of the United States showing the estimated percent of the population ill based on a simulated pandemic influenza model originating in Chicago, IL. Pandemic spread on day 37 with no Pandemic spread on day 37 with all decision measures implemented decision measures implemented
  • 19. Visual Analytics For Syndromic Surveillance• Syndromic surveillance: detection of adverse health events focusing on pre-diagnosis information to improve response time• Pre-diagnosis information can consist of multiple data sources: – Over the counter medicine sales – News reports on emerging diseases – Pro-med news feeds – Emergency department chief complaints April 2012
  • 20. Hypothesis Generation and Exploration Best Paper Nominee, IEEE Symposium on Visual Analytics Science and Technology (VAST), October 2008, for “Understanding Syndromic Hotspots – A Visual Analytics Approach,” 21 (Maciejewski, R., Rudolph, S., Hafen, R., Abusalah, A., Yakout, M., Ouzzani, M., Cleveland, W., Grannis, S., Wade, April 2012 M Ebert D )
  • 21. Time Series Modeling• Seasonal-Trend Decomposition Based on Loess • Time series can be viewed as the sum of multiple trend components • For each data signal, components are extracted • Can then analyze correlation between components Hafen, R., Anderson, D., Cleveland, W., Maciejewski, R., Ebert, D., Abusalah, A., Yakout, M., Ouzzani, M., Grannis, S., “STL for Modeling, Visualizing, and Monitoring Disease Counts,” BMC Medical Informatics and Decision Making, 9(21), 2009. 22 April 2012
  • 22. Predictive Visual Analytics April 2012
  • 23. Advanced Decision Support Tools: RiftValley Fever April 2012
  • 24. Modeling a Pandemic• Pandemic Influenza Planning Tool• Models user specified: • Pandemic influenza characteristics • County population, demographics, hospital beds• Decision measures Map of the United States showing the estimated percent of the population ill • Strategic National based on a simulated pandemic influenza model originating in Chicago, IL. Stockpile deployment • School Closures • Media Alerts Pandemic spread on day 37 with no Pandemic spread on day 37 with all decision measures implemented April 2012 decision measures implemented
  • 25. Visual Analytics Usesfor Public Safety • Risk visualization and analysis • Predictive analytics • Uncertain decision making • Alternative evaluation and consequence investigation • Trend analysis, clustering, anomaly detection • Interactive, multi-day, month, type investigation • Multisource, multimedia data integration & analysis April 2012
  • 26. Example Maritime Public Safety Projects• Search and rescue resource allocation• Swimmer death analysis• PWCS analysis• Economic impact analysis• Resource allocation and risk-based decision making April 2012
  • 27. Example: Visual Analytics Environment• Supports decision making and risk assessment• Interactive exploration and analysis of trends, patterns and anomalies• Allows analysis of risks associated with • Closing one or more Coast Guard stations • Find optimal stations that absorb work load of the closing station • Allocating new resources • Impact on safety and efficiency of operation• Currently being used by analysts at the U.S. Ninth District, HQ, and Atlantic Commands April 2012
  • 28. Example: USCG D9 SearchAnd Rescue Operational Analysis• Interactive visual analytics of multivariate performance metrics for each unit’s activities• Interactive linked spatial temporal display, calendar view, and timeline views April 2012
  • 29. System Features: SAR Risk Profile Time taken by CG stations to deploy an asset to the Great Lakes to respond to a SAR incident. April 2012
  • 30. U.S. Coast Guard Search and Rescue VA (cgSARVA) Partners: USCG LANT 7 (Operational Analysis) , USCG D9, USCG D5IMPACTS:• Analyzed impact of CG auxiliary stations on search and rescue mission in Great Lakes• Used for resource allocation for SAR• Provided evidence of temporal and spatial patterns used in planning – new insights to SAR mission• Hurricane Irene resource allocation decision based on cgSARva analysis and visualization • Highest SAR workload that weekend for D9 April 2012
  • 31. 180 FY  160 2006 140U.S. Coast Guard 120 FY  2007Swimmer Death Analysis 100 FY  2008 80 FY  2009 60Impact: 40 FY  2010• Analyzed spatial and temporal 20 FY  patterns of shore-based and 0 2011 boat-based swimmer deaths to 5 YEAR  AVG understand death dramatic increase in D9 in Summer 2010 Findings:• Provided information and • Swimmer deaths • August highest frequency visualizations used for public • Late afternoon highest frequency information campaign 2011 and for • Lake Michigan (south and west shore) have high concentration patrols 2011 • Boating deaths• Significant decrease in deaths in 2011 • Fri, Sat, Sun account for almost all deaths • Mid July to Mid August have highest frequency Swimmer Deaths – Type ‘Swim’- Apr-Oct 2010 (Lives Lost (only 1 week significantly high)141312 Field)1110 9 8 • 2009-2010 from MISLE Data 7 6 5 4 • Large increase on Mon, Thu, Fri, Sun 3 2 1 0 • Early and late season increase 6 am 6 pm 12 am 10 am 11 am 12 pm 10 pm 11 pm 1 am 2 am 3 am 4 am 5 am 7 am 8 am 9 am 1 pm 2 pm 3 pm 4 pm 5 pm 7 pm 8 pm 9 pm April 2012
  • 32. Resource Allocation and Risk-BasedDecision Making• Explore risk-based decision making and utilize historical data for analysis and prediction • Total Risk, Mitigated Risk, Residual Risk • Explore 11 different USCG missions • Explore allocation of assets with different capabilities • Explore staffing, utilization, assets vs. risk measures • Perform What-If scenarios April 2012
  • 33. Example Public Safety Projects:• Evacuation planning and management• Hazardous Materials - mobile imaging• Law enforcement visual analytics• Additional projects • GARI – gang graffiti • Jigsaw April 2012
  • 34. Large Scale Evacuation Model:Decision Support (Ribarsky et al. UNCC) April 2012
  • 35. Mobile Interface April 2012
  • 36. MERGE: Mobile Emergency ResponseGuide – System Overview Image Capture Image Analysis 1 2 Display Information Database Query 4 3 April 2012
  • 37. MERGE – iVALET Interactive PlumeVisualization and Evacuation Planning• Chemical release plume modeling identifies census tracts with the highest number of expected people affected April 2012
  • 38. Geovisual Analytics Support For SensemakingIn Public Health And Crisis Management(GeoVista Center, PSU) • SensePlace – leveraging news to support infectious disease modeling and delivery of services in a developing country context • SensePlace2 – leveraging social media to support situational awareness for health and crises April 2012
  • 39. SensePlace2: Place-Time-Concept Analytics query time filter & grid map of tweet frequency matching query; window freq display graduated circles depict 500 most relevant tweets; support for spatial filter by distance from point history view ranked, sortable 500 selected most relevant tweets placehttp://www.geovista.psu.edu/SensePlace2/ April 2012
  • 40. Scatterblogs: Geo-Spatial Document Analysis41 April 2012
  • 41. Spatiotemporal Social Media Analytics forAbnormal Event Detection April 2012
  • 42. Visual Analytics Law Enforcement Toolkit(VALET, iVALET)End-Users: Lafayette, WL Police, Indiana Fusion Center, Indiana State Police, Ohio Fusion Center (initiated)Impacts:• In use to analyze crime patterns in Lafayette, Indiana and connect strings of activities• Mobile version being released to public (November 2011) for community-based policing• Investigating correlation of bus routes and crime, street lights and crime• Analyzing time of day problems and improving accuracy of police record VALET delivered: management system • Spring 2011: WL, Lafayette Police• Novel statistical predictive model iVALET delivered: incorporated for planning • October 2011: Purdue, WL Police April 2012
  • 43. Linked Predictive Crime Models by Type Visual Analytics Law Enforcement Toolkit (VALET)- Tippecanoe County Example • Day vs. Night Thefts2008 (red) vs. 2007 (blue background)Aggravated Theft April 2012
  • 44. Drunkenness / Public Intoxication Day-of-the-Week 60 50 40 30 20 10 0 Sat Wed Mon Tue Thu Fri Sun April 2012
  • 45. Football seasonDrunkenness / Public Intoxication Home Away PU vs. Notre Dame PU vs. Iowa PU Lost: 10-38 Homecoming (Sat.) PU Lost: 21-31 PU vs. Illinois PU Won: 21-14 Day-of-the-Week 60 50 40 30 20 10 0 Sat Mon Tue Wed Thu Fri Sun April 2012
  • 46. iVALET • Explore criminal, traffic and civil data on-the-go • Risk assessment • Use current spatial + temporal context into analysis April 2012
  • 47. iVALET • Linked views to explore multivariate spatiotemporal dataset • Analytical tools to help explore data April 2012
  • 48. GeoVISTA CrimeViz… an extensible web-basedgeovisualization applicationthat supports exploration of &sensemaking about criminalactivity in space & timeProject Personnel: Alan M. MacEachren Robert E. Roth GeoVISTA Center Department of Kevin S. Ross Geography Scott Pezanowski Penn State University maceachren@psu.edu http://www.geovista.psu.edu/CrimeViz April 2012
  • 49. GeoVISTA CrimeViz: Year/Month View Aggrevated Assault, by month, for a year; map depicts freq/cell in Nov. April 2012
  • 50. Crime Analytics & SA Issues and Techniques• Fuse data from a variety of sources • Law enforcement records management • Weather and phases of the moon • Street light locations, bus routes • Tracking release data of offenders • Civil court data • Social Media • Local event calendar• Reliable predictive models• Understandability and trust of predictions• Main Question: What helps officers, detectives, chief do their variety of jobs? April 2012
  • 51. Jigsaw:Visual Analytics for Investigative Analysisand Exploration of Document Collections “Putting the pieces together” Goal: Assist investigators with understanding, sense-making, and analysis of large, unstructured and structured document collections Approach: Provide multiple visual perspectives on the documents and entities within them, highlighting connections between entities April 2012
  • 52. Gang Graffiti Recognition and Analysis Using a Mobile Telephone (GARI)Edward J. Delp, Mireille BoutinCollaborating Institution(s): Purdue UniversityEnd-User(s): Gary Coons, Chief/Indianapolis Department of Public Safety Division of Homeland Security andIMPD, Indiana Fusion Center Gang Task ForceIMPACT: • Allows police to catalog and analyze gang GARI delivered: graffiti images into a database system to better • Summer 2011: track and determine gang activity throughout a • IMPD gang detectives • August 2011: region • IMPD at large • Will allow the graffiti images to be “interpreted” • Ind Fusion Gang Task Force • More than 40 users and 450 graffiti images • October 2011: acquired • Gang detectives across Indiana April 2012
  • 53. PathIntroduction to our Center - VACCINEMotivation and introduction to visual analyticsExample solutions we’ve developed Health visual analytics Public safety Resource allocation Visual analytic challenges and future April 2012
  • 54. Five Six Challenges for Proactive & Predictive Visual Analytics1. Creating computer-human visual cognition environments2. Integrating interactive simulation and analytics3. Solving specific scale issues and cross-scale issues4. Managing uncertainty and time5. Enabling risk-based decision making environments6. Developing the Science of Interaction for Visual Analytics April 2012
  • 55. Visual Analytics At Real-World Scale• Utilize advanced HPC • Example: Longhorn techniques to enable Exascale Visual Analytic interactive spatiotemporal Platform analysis (spatiotemporal • 2048 compute cores (Nehalem clustering, prediction) quad-core)• Cluster-based and cloud- • 512 GPUs (128 NVIDIA Quadro based solutions Plex S4s, each containing 4 NVIDIA FX 5800s)• GPGPU solutions • 13.5 TB of distributed memory• Develop easily usable • 210 TB global file system HPC visual analytic environments April 2012
  • 56. The Science of Interaction• Definition: The study of methods by which humans create knowledge through the manipulation of an interface. • …interaction and inquiry are inextricable. It is through the interactive manipulation of a visual interface – the analytic discourse – that knowledge is constructed, tested, refined and shared.” Stasko et al., IV20095 Challenges:1.Visual Discourse - Interactive visual thinking tools for the exploration, understanding, collaboration, description, explanation, decision, dissemination, persuasion of concepts & data.2.Multi-modal Sensemaking In The Large - Enable large-scale, distributed teams to interactively make sense of big multi-modal data and problems.3.Fluid Interaction - Designing fluid, high-bandwidth, and powerful interaction models and paradigms for the purpose of individuals reaching their full potential in viewing, analyzing, and understanding large and complex data.4. Collaboration – beyond space and time- synergizing technologies and human users5. Mixed Initiative Data Discovery and Manipulation - Balancing active user input with systematic guidance to enable visual data manipulation and analysis. April 2012
  • 57. Visual Analytics: Remember…• We need to be cognizant of parameters for visual representations• Appropriate analysis can guide users to interesting features in the data• Refined analysis through user interaction and their domain knowledge can help discover hidden problems• There is no single catch-all visual representation or analysis April 2012
  • 58. Keys for Success• User and problem driven• Balance human cognition and automated analysis and modeling • Often applied on-the-fly for specific components identified by the user• Interactivity and easy interaction • Utilizing HPC and novel analysis approaches• Understandability of why predicted value is what it is• Intuitive visual cognition• Not overloaded with features April 2012
  • 59. For Further Information www.VisualAnalytics-CCI.orgvaccine@purdue.eduebertd@purdue.edu April 2012

×