Information Visualization and Visual Analytics
Pekka Wartiainen
University of Jyv¨askyl¨a pekka.wartiainen@jyu.fi
23.4.2014 Outline
Objectives
Introduction
Visual Analytics
Information Visualization
Our research
Summary Learning objectives
I To understand the definition of visual analytics.
I To be aware with visual analytics approach in problem solving.
I To understand the basics of data visualization. Motivation
I Raw data has no value in itself, only the extracted information has value
I Time and money are wasted and opportunities are lost
I Success depends on availability of the right information
I Visual analytics aims at making data and information processing transparent
I Visual analytics combines the strengths of humans and computers An historical perspective on visual analytics
I Early visual analytics: exploratory data analysis
I Visual data exploration and visual data mining
I First book of visual analytics: Illuminating the Path, 2005
I Some earlier systems exhibited the characteristics of visual analytics
I CoCo system for improving silicon chips, 1990 Past few years
I VisMaster is an European Coordination Action Project
I Web-page:
I URL: Visual-Analytics.EU
I Book:
I URL: Mastering the information age - solving problems with visual analytics
I YouTube video:
I URL: Inria - Vismaster, visual analytics Visual analytics
Definition Visual analytics combines automated analysis techniques with interactive visualisations for an effective understanding, reasoning and decision making on the basis of very large and complex datasets. Timeline Application of visual analytics
I First application area was security
I Many major application areas
I physics, astronomy, medicine, climate, . . .
Example: business intelligence
I Financial market generates large amounts of data on a daily basis –> extremely high data volumes over the years
I More than 300 million VISA credit card transactions per day
I Multiple perspectives and assumptions for analysis
I history, current situation, monitoring, forecasting, recurring situations Visual analytics – Coordinated Graph Visualization
Visual support for the simulation of climate models provided by CGV, a highly interactive graph visualization system. Visual analytics – NFlowVis
Analysis of a distributed network attack. The visual analytics process
Process model of visual analytics. Building blocks of visual analytics research
Visual analytics integrates science and technology from many disciplines. Evaluation
I Evaluation include techniques, methods, modes and theories as well as software tools
I Challenge: often processing data from the real world
I Evaluation involves users, tasks and data
I Especially in the industry, the domain expert has the best knowledge –> Empirical evaluation
I Evaluation criteria, e.g.:
I effectiveness I efficiency I user satisfaction
I Importance of documentation is emphasized Infrastructure
I Visual analytics is both user-driven and data-driven
I Current challenges: lack of interaction and dynamic data
I Limitations of traditional data bases
I Old fashioned ‘architectural reference model’ I Big data solutions
I Need for:
I Fast imprecise answers with progressive refinement I Incremental re-computation, either in the data (e.g., some data has been changed) or in the analysis parameters I Steering the computation towards data regions that are of higher interest to the user. Data management – Why?
I The big opportunity of the Information Age
I Many obstacles need to be overcome
I Heterogeneity of data sources I Different data types I Data streams I Working under pressure I Time consuming activities
I Data management ensures data consistency and standards Data management – VA aspects
I Data and semantic integration
I Utilizing known processing methods
I Data Warehousing, OLAP and Data Mining
I Data reduction and abstraction
I Data quality is crucial (cf. GIGO model)
I Visual techniques for exploring data Space and time
I In large systems, space and time are essential –> complexity increases
I Space and time are more than just numbers
I Specific properties:
I Dependencies between observations I Uncertainty I Scale I Time
I Spatial approaches: Cartography, GIS, Geovisualization
I Representation of time: visualization of time-related data and time itself
I Interactive visualizations
I Big data cases – dimension reduction Space and time – OECD eXplorer
Allows to explore regional statistics data from OECD URL:Organisation for Economic Cooperation and Development Data mining
I Humans are required in the data analysis process
I New tools and methodologies are necessary to help experts extract relevant information
I Limitations in KDD process and visualizations
I Combination of multidisciplinary approaches
I Pattern identification methods
I Spatio-temporal data mining
I Many software have been developed Perception and cognitive aspects – visualization
I The human is at the heart of visual analytics human interaction, analysis, intuition, problem solving and visual perception.
I Distinction between high and low-level vision
I Humans do not have to remember everything but extract visual clues from the environment
Pre-attentive processing makes items pop out the display automatically. Data visualization
I Fast and understandable way to present data to a user
I Data mining methods as pre-processing tools
I Many visualization methods existing
I JFreeChart I Google Charts
I Remember how not to use visualization techniques
I Dynamic behavior of the data sets special requirements
I Data visualization is part of information visualization GUI design
I Visual analytics has high demand for GUI
I Scalable and interactive interface
I General guidelines for different purposes
I Windows, OS X, Android, . . . I Online solutions
I Define target group before designing the GUI
I Multidisciplinary research groups I Personalized user roles Common interaction
I select : mark data items of interest, possible followed by another operation,
I explore : show some other data e.g., panning, zoom, resampling,
I reconfigure : rearrange the data spatially e.g., sort, change attribute assigned to axis, rotate (3D), slide,
I encode : change visual appearance e.g., change type of representation (view), adjust colour/size/shape,
I abstract/elaborate : show more or less detail e.g., details on demand, tooltips, geometric zoom,
I filter : select or show data matching certain conditions,
I connect : highlight related data items e.g., brushing (selection shown in multiple views). Using colors
I Powerful element in visualization
I Wrong usage of colors is disturbing
I Color Usage Research Lab
I NASA Ames research center
I Ready made color palettes are solid alternatives Visual analytics in energy production
I Application area: BFB boiler burning biomass
I Co-operation with VTT, department of chemistry, and private companies
I Funded by Regional Council of Central Finland
I Time-series data measured from the different parts of the process
I Context-sensitive framework approach
I Matlab routines with Java GUI People included into process
The human context of visual analytics. Summary
I Visual analytics for multidisciplinary research problems
I Visualization, data analysis, user interaction
I Highly interactive interfaces
I The whole process should be taken into account
I Many challenges still existing, especially with big and dynamic data
I Humans are part of the process References
D. Keim, J. Kohlhammer, G. Ellis ja F. Mansmann, Mastering the Information Age: Solving Problems with Visual Analytics, Eurographics Association, Germany, 2010.
P. J¨arvinen,K. Puolam¨aki,P. Siltanen ja M. Yliker¨al¨a,Visual Analytics, Technical report, VTT, Finland, 2009.
P. Wartiainen, T. K¨arkk¨ainen,A. Heimb¨urger,ja S. Ayr¨am¨o.¨ Context-sensitive approach to dynamic visual analytics of energy production processes. In 22th European-Japanese Conference on Information Modelling and Knowledge Bases. MATFYZPRESS - Univerzity Karlovy, 2012.