Introduction to Information Visualization

Preliminaries Intro Vis Process
Introduction to Information Visualization
Alexander Hinneburg
14.9.2004
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Dates of the Lectures
Week
38
39
40
41
42
43
Thuesday
14.9.
21.9.
28.9.
5.10.
12.10.
19.10.
Thursday
16.9.
23.9.
30.9.
7.10.
14.10.
21.10.
Time and Place: 16-18 o’clock in room C222, Exactum
Office Hours: Wednesdays 13-14 o’clock, Room A337, Exactum
Homework: will be discussed the next week in the last part of
the Thuesday lecture
Projectwork: will be given in the last third of the lecture series
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Structure of the Lecture (2)
I
Introduction
I
I
I
I
I
I
I
Goals of Visualization
Requierements
Expressiveness
Effectivity
Costs
Examples
Visualization process
I
I
I
The visualization pipeline
A reference model for visualization
Visualization scenarios
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Structure of the Lecture (2)
I
Visual Perception
I
I
I
Visual acuity and contrast sensitivity
Color, texture, shape and 3D-depth
Temporal change and motion
I
Basic techniques
I
Advanced visualization of information
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
What is Visualization?
I
Visualization means literally:
bringing something into the presence, makeing something
visible
I
Visualizations serve the communication of concents
I
Visualization is the transformation of symbolic contents into
geometric contents [MFB87]
I
“The medium is the message” [McL64]
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
What is Visualization?
I
I
Visualization is one of the oldest communication media
It has been used in:
I
I
Arts, like painting, for religious and secular subjects
Astronomy, meteorology, cartography
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Chinese Map, 11th century AD
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
European Map, 1546 AD
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Cholera Epidemic, London, John Snow, 1854
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
What is Visualization?
Visualization in Computer Science:
I
Presentation of data
I
Translation and analysis of data
I
Detection of hidden coherences
I
Avoidance of wrong coherences
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Statistical Diagrams (1)
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Statistical Diagrams (2)
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Statistical Diagrams (3)
I
Look at the lables of the plots.
I
Does it makes sense to you?
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
The Goal of Visualization
I
The beholder shall develop a mental model by viewing the
visualization. The visual attributes have to match the
properties of the data in a well defined manner.
I
With the help of the visualization, the beholder shall be able
to reconstruct its context in the real world and also he/she
shall be able to match the recognized structures with the real
world.
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
The Goal of Visualization (2)
A visualization shall ease
I
the analysis, the understanding and the communication of
I
Models, concepts and data
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Usage of Visualization
I
Explorative Analysis
I
I
I
Confirmative Analysis
I
I
I
no Hypothesis available, take only the data as basic
interactive, undirected search for information and structures
additionally to the data hypothesis are avaiable
Goal to check and verify those hypothesis
Presentation and Communication
I
I
I
The author have to take the target group into account
Facts and Statements are to be presented to people not so
familiar with your stuff
The shown data shall be understood and interpreted without
big problems
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
General Requirements
I
I
Ultimate goal: make hidden coherences visible to non-experts
Visual levels of Information [Ber82]:
I
I
I
elementary level: all explizit facts of the data are represented in
the visalization
medium level: shows also abstractions of the basis information,
goal: Communications
high level: shows all explizit facts as well as the hidden
coherences at large, so it can serve as basis for furhter
decisions, such a visualization comprises also the lower levels.
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Quality of Visualizations
I
Defined by the degree the communication goal is reached
I
The ratio of the amount of information recognized by the
beholder compared to the amount of information, which s/he
undertstood in that timeframe
I
The quality depends from the data, working goal, properties
of the used media, von den capazities and experiences of the
beholder
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Napoleons campaign against Russia
Charles Joseph Minard [1781-1870]
How many variables are shown?
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Quality of Visualizations (2)
Questions from [Rob91]
I
Which mental model represents the different kinds of
information most effective and is most suitable for
communication?
I
Which defined and recognizeable visual representation is most
suitable to show some specific information?
I
How can the chosen model be made accessible to the
beholder in the most effective way?
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Influential Factors
I
Type and structure of the data
I
Working goal of the visualization
I
Domain knowledge of the user
I
Visual capabilities or preferences of the beholder
I
Common metaphors and conventations in the domain
I
Characteristics of the media
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
More specific criteria
A visualization has to be
I
expressiv
I
effective and
I
appropriate
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Expressiveness
The second figure is not expressive as it suggests a qualitative
rating of the countries which is not intended.
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Expressiveness (2)
How to show that there is nothing?
„The urge to see"´, Joseph Koudelka, Prag, 22.8.1968
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Effectivity
I
Effectivity depends not only on the data but also:
I
I
I
on the working goal and
on the capabilities of the beholder
An effective visualization trys to present the contents in a
rather intuitive way
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Effectivity
Both figures show house prices, however only the right hand one is
easy to read, because the relation between prices at different
locations can be compared directly without looking at the legend.
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Effectivity (2)
Goal: find histograms with more than one peak
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Effectivity (3)
Goal: find histograms with more than one peak
Histograms are color coded bars, dark means high bin value
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Effectivity (4)
Goal: find histograms with more than one peak
Histograms are color coded bars, dark means high bin value
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Appropriateness
I
Expressiveness and Effectivity are necessary conditions for a
visualization which take the beholder into account. However,
they do not measure the costs for generating the visualization.
I
Appropriateness describes the amount of resources used to
create a visualization.
I
Effectivity and Appropriateness are often strongly coupled in
practice.
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
The Visualization Process
Content
I
The visualization pipeline
I
A reference model for visualization
I
Visualization scenarios
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
The Visualization Pipeline
I
Data preparation (Filtering)
I
Transformation into geometric information (Mapping)
I
Image generation (Rendering)
Data
- Filtering
- Mapping
Alexander Hinneburg
- Rendering
Information Visualization
- Image
Preliminaries Intro Vis Process
Data preparation
I
I
Realizes a data to data mapping
Possible opertations:
I
I
I
I
I
Completion, Interpolation
Projection (reduce no. of variables)
Selection (apply criterias, smoothing, cut off operations)
Determine implicit properties (max, gradient)
Conversion
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Mapping
I
The prepared data are transformed into a geometric model
I
Select geometric primitives and assign the data attributes
I
Make sure that expressiveness and effectiveness criteria are
met
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Rendering
I
I
Geometric data are transformed into an image, different
graphic packages support this step
Different ways of rending are possible:
I
I
I
I
Images close to reality
Abstract images
“Mental” images
Animations
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Visualization Pipeline
Raw
Data
- Prepared
Data
- Geometric
Data
- Image Data
(a) Points of measurement, (b) Interpolation onto a grid,
(c) generating a geometric model and rendering an image
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Data Selection
I
Selection or focusing on interesting data may be applied at
different position in the visualization pipeline
I
I
I
I
Selection of raw data
Selection of prepared data
Selection of geometric data
Focussing on some part of the image
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Distribution of the Pipeline
I
The publisher generates an image
I
I
I
I
I
Publisher performs all steps of the vis pipeline
Viewer has no possibility for interaction
Publisher send the image to all viewers, high bandwidth
required
Consistency with the data may be a problem
The publisher generates a geometric model
I
I
I
Viewer renders the image
Viewer has some possibilities of interaction, e.g. walk through
Examples are Java applets or VRML plug-ins
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Distribution of the Pipeline (2)
I
The publisher delivers the raw data and the viewer generates
the visualization
I
I
I
I
Viewer has a lot of freedom
Expertise and tools are necessary at the viewers side
Data formats might be a problem
The publisher generates a geometric model under the control
of the viewer, the viewer takes care of the rest of the pipeline
I
I
Compromise to eliminate disadavantages of the previous
scenarios
Special interface needed for the comunication between viewer
and publisher
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Reference Model [RF94]
How to incorporate the requirements for expressiveness and
effectivity into the visualization pipeline?
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Cycle of the visual Analysis
Interface to the
Visualization
Observation
Measurements
Modelling
Data
Visualization
(Filtering,
Mapping
Rendering)
Simulation
Visualization Scenarios:
I Video Mode
I Tracking Mode
I Interactive Postprocessing
I Interactive Control
Alexander Hinneburg
Information Visualization
Visual Analysis
Preliminaries Intro Vis Process
Video Mode
1. Step
2. Step
Observation
Measurements
Modelling
3. Step
Data
Video
Data
Simulation
Visualization
(Filtering,
Mapping
Rendering)
Visual Analysis
Video
I
Pro: No time limit for the generation
I
Con: no interaction is possible
I
Well suited for molecul simulations, flow fields, ...
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Tracking Mode
Observation
Measurements
Visualization
(Filtering,
Mapping
Rendering)
Modelling
Visual Analysis
Simulation
I
Direct coupling of visualization to the data generating
processes
I
Pro: good understanding of partial steps
I
Con: High requirements for the hardware
no interaction
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
Interactive Postprocessing
1. Step
2. Step
Observation
Measurements
Modelling
Data
Data
Visualization
(Filtering,
Mapping
Rendering)
Simulation
I
Often used in practice
I
Separation of data generation and visualization
I
Postprocessing of the data is often very useful
Alexander Hinneburg
Information Visualization
Visual Analysis
Preliminaries Intro Vis Process
Interactive Control
Observation
Measurements
Visualization
(Filtering,
Mapping
Rendering)
Modelling
Visual Analysis
Simulation
I
Integration of visualization and data generating processes
I
The user can influence bside the visualization also the
modelling and simulation part
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
[Ber82]
J. Bertin.
Graphical Presentations and the graphical processing of
information.
de Gruyter, New York, 1982.
[McL64] Herbert Marshall McLuhan.
Understanding media, the extension of man.
MIT Press, 1964.
[MFB87] B. H. McCormick, T. A. De Fanti, and M. D. Brown.
Visualizations in scientific computing.
Computer Graphics, 21(6):1–14, 1987.
[RF94]
G. Robertson and L. De Ferrari.
Systematic approaches to visualization: is a reference model
needed.
In Scientific Visualization, pages 287–305. Academic Press,
1994.
Alexander Hinneburg
Information Visualization
Preliminaries Intro Vis Process
[Rob91] P. K. Robertson.
A methodology for choosing data representations.
IEEE Computer Graphics and Applications, 11:56–57, 1991.
Alexander Hinneburg
Information Visualization