Information Visualization & Visual Analytics Jack van Wijk - - PowerPoint PPT Presentation

Information Visualization & Visual Analytics

Jack van Wijk

• Dept. Math. & Computer Science

TU Eindhoven BPM round table, March 28, 2011

Overview

• InfoVis
• Visual Analytics

Why is my hard disk full?

?

SequoiaView

• www.win.tue.nl/sequoiaview

Van Wijk et al., 1999, Bruls et al. 2000

Information Visualization

• The use of computer-supported,

interactive, visual representations of abstract data to amplify cognition (Card et al., 1999)

Information Visualizatio n Abstract dataset (table, graph, tree) User image data interaction

Abstract data

• Multivariate data visualization
• Tree visualization
• Graph visualization

scatterplot tree diagram node link diagram

InfoVis at TU/e

Focus:

• Large data sets, professional users
• Use of computer graphics know-how

– shading, geometry, texture, …

• Software Visualization

(similar issues as BPM?)

Software Visualization

• User: developer, architect, manager, …
• Some fuzzy questions:

– Is the structure sound? – Can I improve the structure by refactoring? – What has happened with the system? – Does the implementation conform the architecture? – Where are the weak spots?

Different views on software

• Architecture

– System structure – Data – Coordination, temporal aspects

• Code

– Structure – Derived data, metrics – Evolution

• Execution

– Traces, call graphs

Challenges in Software Visualization

Combination of large amounts of

– Multivariate data (metrics) – Hierarchical data (system, subsystem, module, ..) – Graph data (call relations) – Text (names, code)

+ + =

Trees + graphs

• Ubiquitous!

MatrixView

Data:

– hierarchy of layers, units, modules, classes, methods – methods calling each other

MatrixView

A C B E D A B C D E A B C D E

Matrix representation of graph

MatrixView

Van Ham 2003, Van Wijk et al., 2003

Hierarchical Edge Bundles

• Again, tree+graph, but now completely

different

Holten, 2006

Showing directions in edges

arrow light-to-dark dark-to-light green-to-red curved tapering

Holten et al., 2009

Result of experiments

Visual Analytics: Beyond visualization

Origin

• Founder: Jim Thomas, NVAC
• Illuminating the Path, 2004

Visual Analytics: The science of analytical reasoning facilitated by interactive visual interfaces

Definition

• The science of analytical reasoning

facilitated by interactive visual interfaces

– Compact! – Complete! – Perfect! – But what is it?

Video

• VisMaster

An InfoVis perspective

Information Visualization Abstract dataset (table, graph, tree) User image data interaction

An InfoVis perspective

Information Visualization Abstract dataset (table, graph, tree) User image data interaction Many, large, heterogenous datasets

• gigabytes, terabytes, petabytes
• tables, images, documents, videos, audio,…

Data mining

• statistics, machine learning,

pattern recognition, artificial intelligence, … Professional

• domain expertise
• fit in workflow
• from data foraging

to presentation

• teamwork

HCI perception cognitive psychology software engineering graphics mathematics design art data management statistics

The key ingredients

• Huge, heterogenous data sets
• Integration of data mining and visualization
• Integration in workflow
• Support for all stages of data analysis
• Support for multiple users
• Keyword: INTEGRATION
• Result = product of parts (2 x 2 x 2 x 2 x 2 = 32)

FAQ

We know this already, isn’t it just:

• applied infoVis, visual data mining, visual data

analysis, statistical graphics, … Sure, Visual Analytics builds on existing technologies and earlier examples exist…

One year of time-series data

Van Wijk et al., 1999 0:00 12:00 24:00 365 graphs #people at work

After clustering

Van Wijk et al., 1999 0:00 12:00 24:00 365 graphs #people at work

Command Post of the Future

• Steven Roth et al.
• Visage (1996), CoMotion, MAYA Viz

Interaction, heterogenous data, knowledge sharing, teamwork, decision making, …

FAQ

We know this already, isn’t it just:

• applied infoVis, visual data mining, visual data

analysis, statistical graphics, … Sure, Visual Analytics builds on existing technologies and earlier examples exist… but integrating all of these is still novel, difficult, and challenging.

FAQ

• This Visual Analytics, that’s American, right?
• No, wrong.

• EU-funded Coordination Action Project
• 26 partners, 12 countries
• Developing roadmap
• Organizing events
• Communication platform
• Video (youtube: vismaster)

Jörn Kohlhammer Daniel Keim

Summary

Visual Analytics:

• Great!
• Big!
• Challenging!