introduction to data visualization - purdue...

9/29/2016

1

Introduction to Data Visualization

Vetria L. Byrd, PhDAssistant Professor

CGT101 Data Visualization Talk

Purdue University

September 29, 2016

Introductions

9/29/2016

2

Academic Preparation

• Computer Science (PhD, MS)

• Biomedical Engineering (MSMBE)

Where I Am NowAcademic Appointment

Assistant Professor

Purdue University

Computer Graphics Technology

Research Focus: Data Visualization

Vetria L. Byrd, PhD

What I’ve DoneVisualization Initiatives

• BPViz: Broaden Participation in Visualization (2014/2016)

• Research Experience for Undergraduates in Collaborative Data Visualization Applications (2014/2015)

Agent for “Insight”

High Level Overview

Purpose Of Visualization

Visualization Applications

Hands-on Introduction using ParaView

AGENDAINTRODUCTION TO DATA VISUALIZATION

9/29/2016

3

7 Billion: How Did we Get so Big so Fast?

Baby names

The Origin of Species

Gender and Teacher Reviews

Wind Map

Prime Number Patterns (pretty cool)

What is Visualization?

• Representing large amounts of disparate information in a visual form often allows you to see patterns that would otherwise be buried in vast, unconnected data sets. …

• Visualizations allow you to understand and process enormous amounts of information quickly because it is all represented in a single image or animation.

What is Visualizationhttp://mindymcadams.com/tojou/2011/10-useful-resources-about-data-visualization/

9/29/2016

4

DATA VISUALIZATIONHTTP://MINDYMCADAMS.COM/TOJOU/2011/10-USEFUL-RESOURCES-ABOUT-DATA-VISUALIZATION/

What is the purpose of Visualization?

9/29/2016

5

“The purpose of visualization

is “insight”,

not pictures.”~Ben Shneiderman

What does Insight lead to?

9/29/2016

6

Visualizing Patterns over Time

“Insight” Leads to . .

Spotting Differences


9/29/2016

7



http://www.kidsfront.com/spot‐differences/9.html

10101010101010101010101010010101010101070101010101010010700101100110011001100110011001110011001100101010101010701010101011100010111000101111000101001101010101010101011100011001010101010101010001010701001010001010101010101010101010101010101010101010101010101010101010101010101010101010101070101011010107010101010101070101001010101010101010101010101001010010110011001100110011001100111001100110010101010101010101010101110001011100010711100010100110101010101010101110001100101010101010101000101010100101000101010101010101070101010101010101010107010101010101010101010101010101010101010101010101110011001100110010101001

• Visualizing Patterns

• Spotting Differences

How many

7’s do you

see?



9/29/2016

8

Allows users to answer questions they didn’t know they had


Human Genome Projecthttps://pradipjntu.files.wordpress.com/2011/05/molecularmachine.jpg

The Challenger Disaster

http://en.wikipedia.org/wiki/33 File: Challenger_explosion.jpg


9/29/2016

9

Visualizing Spatial Relationships


Muehlenhaus, I. (2012). Chapter 8, Visualizing Spatial Relationships, Visualize This: The Flowing Data Guide to Design, Visualization, and Statistics, pp 271‐326.

http://datafl.ws/197

http://datafl.ws/198

Watch the Growth of Walmart and Sam's Club

Watch the Growth of Target Stores

Can anyone think of another reason why “insight” is important?

9/29/2016

10

“Insight” Tells a Story.

Temperature

Army Size: 422,000

Army Size: 10,000

Army Size: 100,000

Moscow

‐11

‐30‐26

‐21

‐9

Path of retreat

Best Statistical Graphic, Ever!

“Insight” Tells a Story

9/29/2016

11

The Joy of Stats - BBC Four

Insight

Explanation

Tells a Story

9/29/2016

12

Why is visualization important?Visualization Applications

Visualization Applications

BioVis

InfoVis

GeoVis

SciVis

The visualization of biological data

“Visual representation” of large‐scale collections of non‐numerical information

Communicates geospatial information

Visualization of three‐dimensional phenomena

9/29/2016

13

Biovisualization(BioVis)

The visualization of biological data;

Often grouped with computer animation

Information Visualization (InfoVis)

Interdisciplinary

Study of the “visual representation of large-scale collections of non-numerical information

InfoVis

Internet UsageSource: http://www.cernea.net/wp‐content/uploads/2013/03/internet.gif

9/29/2016

14

Geographic Visualization

GeoVis

Communicates geospatial information in ways that, when combined with human understanding, allow for data exploration and decision‐making processes.

MacEachren, A.M. and Kraak, M.J. 1997 Exploratory cartographic visualization: advancing the agenda. Computers & Geosciences, 23(4), pp. 335‐343. Jiang, B., and Li, Z. 2005. Editorial: Geovisualization: Design, Enhanced Visual Tools and Applications. The Cartographic Journal, 42(1), pp. 3‐4 MacEachren, A.M. 2004. Geovisualization for knowledge construction and decision support. IEEE computer graphics and applications, 24(1), pp.13‐17

Scientific Visualization(SciVis)

Primarily concerned with the visualization of three-dimensional phenomena

Emphases on realistic renderings of volumes, surfaces, illumination sources, etc.

http://www.tinkering.net/sciviz/

Image Source: http://www.sci.utah.edu/the‐institute/highlights/24‐research‐highlights/cibc‐highlights/253‐top‐scientific‐visualization‐research‐problems.html

9/29/2016

15

Taking raw data and converting it to a form that is viewable and understandable to humans.

Adopted from The ParaView Tutorial, The Basics of Visualization, version 3.98

There are several steps between raw data and a finished visualization


9/29/2016

16

Why do we care?I just want a pretty picture!

9/29/2016

17

• Multi-platform parallel data analysis and visualization application

• Mature, feature-rich interface

• Good for general purpose, rapid visualization

Mac

Windows

Linux

9/29/2016

18

• Open Source . . . It’s Free!

• http://www.paraview.org/

• Built upon the Visualization Toolkit (VTK) library

• Primary contributors:

Kitware, Inc.

Sandia National Laboratory

Los Alamos National Laboratory

Army Research Laboratory

Grid – regular structure, all voxels (cells) are the same size and shape


9/29/2016

19

Curvilinear – regularly gridded mesh shaping function applied


Unstructured grid – irregular mesh typically composed of tetrahedra, prisms, pyramids, or hexahedra


9/29/2016

20

• Point data

• Polygonal data

• Images

• Multi-block

• Adaptive Mesh Refinement (AMR)

• Time series support

• Isosurfaces

• Cutting planes

• Streamlines

• Glyphs

• Volume rendering

• Clipping

• Height maps

• & more

9/29/2016

21

• WRF weather forecast data set

Rectilinear grid

Multiple scalar and vector variables

Time series

• Can show:

Clouds

Wind

Temperature

Image created by Paul Navratil and Kelly Gaither (TACC)

Three Basic Steps:

• First your data must be read into ParaView

• Next, you may apply any number of filters that process the data to generate, extract, or derive features from the data

• Finally, a viewable image is rendered from the data

9/29/2016

22

ParaView 5.0.0Let’s get started . . . .

Sanity Check

Software Installed? Data Sets downloaded? Can you locate the datasets?

Data set: headsq.vti

9/29/2016

23

Training Data

Download the data file

http://web.ics.purdue.edu/~vbyrd/trainingData.html

• Right‐click on headsq.vti• Save the file on the desktop

Load Data File

Extract Isosurfaces from the data

Create contours, clip contours, slice contours

Create glyphs, tubes, streamlines

Edit Color Map

Volume Rendering

Saving your Data

Getting your data into Paraview

Additional Resources

Contour

Clip

9/29/2016

24

Menu Bar

Tool Bar

Pipeline Browser

Object Inspector

3D Viewer

Let’s try visualizing a data setheadsq.vti

9/29/2016

25

FileOpen

headsq.vti

New Object in Pipeline Browser

Click Apply

9/29/2016

26

Should see a bounding box in the 3D viewer window

Create an Isosurface

Select: Filters Common Contour

9/29/2016

27

A new object appeared in the pipeline browser(Contour 1)

Contour – Extracts the points, curves, or surfaces where a scalar field is equal to a user‐defined value.

The surface is often also called an isosurface.

Value Range for the data set is now visible

9/29/2016

28

Value Range for the data set is [ 0, 4095 ]

Only one value is showing: 2047.5

Click Apply to see what points, curves, or surfaces in the dataset have a value of 2047.5

If you do not see anything in the 3D window click the eye icon next to Contour1 in the Pipeline Browser

This allows you to toggle between views in the 3D Viewer

9/29/2016

29

+X ‐X +Y

‐Y +Z ‐ Z

9/29/2016

30

You should be here

Pipeline Browser Two objects Value Range

[0, 4095]

Select 2047.5 showing in Value Range

Delete that value (click the minus button to remove all values)

9/29/2016

31

Click the button below the minus button to: Add a Range Of Values

Should see the Add Range Window Use this window to set the

range of values For this tutorial

Min: 0 Max: 4095

Feel free to pay around with the range (between 0 and 4905)

Click OK

Notice the Value Range: [0, 4095]

There are 10 values (steps) showing values between 0 and 4095

9/29/2016

32

This may take a few seconds to render . . .

9/29/2016

33

Contour – Extracts the points, curves, or surfaces where a scalar field is equal to a user‐defined value.

The surface is often also called an isosurface.

9/29/2016

34

CLIP ‐ Intersects the geometry with a half space.

The effect is to remove all the geometry on one side of a user‐defined plane.

Select:Contour 1 (in

pipeline Browser)

FiltersCommonClip

1

2

34

9/29/2016

35

A new object appeared in the pipeline browser(Clip 1)Clipping plane (see red vertical line and horizontal arrow)

Select the arrow point (arrow turns red)

Rotate (drag) the arrow point until the arrow is pointing out of the screen toward you)

9/29/2016

36

9/29/2016

37

Depending on where you placed the clipping plane the results may be easily seen: see clipped ears; and area round neck

Make sure the eye icon is not greyed out on the Clip1 object in the Pipeline Browser

Rotating the view reveals the clipped isosurface

9/29/2016

38

If this property is set to 0, then clip filter will return that portion of the dataset that lies within the clip function.

If set to 1, the portions of the dataset that lie outside the clip function will be returned instead

9/29/2016

39

Click the eye icon next to Clip1 in the pipeline browser (hide the clip plot)Select Contour 1

Select Filters Common Slice

9/29/2016

40

Rotated View

Drag arrow point around to point out of screen toward you.

9/29/2016

41

SLICE – Intersects the geometry with a plane. The effect is similar to clipping except that all that remains is the geometry where the plane is located.

9/29/2016

42

headsq.vtiheadsq.vti

Contour1

Clip1 Slice1

headsq.vtiheadsq.vti

Q: How do we combine

(show) the Clip and Slice views at the same time?

+

9/29/2016

43

Let’s try visualizing a data setExploring the range of values in the data set

9/29/2016

44

Step 1: Delete the existing range of values

Step 2: Add a new range of values

From To Step What do you get?

1500 2500 10

2500 3500 10

2999 3500 20

3500 4000 20

2000 2999 10

3000 3999 10

Try different Value Ranges

Don’t ForgetDon’t Forget

9/29/2016

45

Exercise: Try parallel rendering of the sample data file on Bluewaters; see connection details below

https://bluewaters.ncsa.illinois.edu/paraview1

https://bluewaters.ncsa.illinois.edu/paraview1

9/29/2016

46

Recall Three Basic Steps:First your data must

be read into ParaViewNext, you may apply

any number of filters that process the data to generate, extract, or derive features from the dataFinally, a viewable

image is rendered from the data

Opened simple data file headsq.vti

Applied filters: Contour, Slice, Clip

Rendered: Output to the 3D Viewer window

Why should you care about Visualization?

9/29/2016

47

• There is a demand for people who understand the visualization process and is able to transform raw complex data into a visual representation that does not overwhelm.

Why Should You Care About Visualization

https://en.wikipedia.org/wiki/Data_science

9/29/2016

48

https://www‐01.ibm.com/software/data/bigdata/images/4‐Vs‐of‐big‐data.jpg


9/29/2016

49



9/29/2016

50

IBM Big Data Platform


21ST CENTURY CHALLENGES

9/29/2016

51

NEW MAJOR FOR UNDERGRADUATES

Visualization LabKNOY 373

Looking for Agents of Insight

9/29/2016

52

BYRD VISUALIZATION LAB Logo Needed Logo

Needed

Research Collaboration

Training Outreach

• Big Data• High Performance

Computing in Visualization

• In‐situ Visualization• Collaborative

Visualization

• Hands‐on workshops

• Invited Talks & Presentations

• Data Visualization Courses

• Broadening Participation and Inclusion

• Collaborative Data Visualization

9/29/2016

53

FUN PROJECT

DEVELOP AN APPLICATION THAT WILL RECOMMEND THE MOST EFFICIENT WAY TO PACK

9/29/2016

54

Given



What does the data look like?

What specs are needed?

Where would you begin?

9/29/2016

55

What are some other applications of this project?

Leap Motion ProjectUsing Gestures

9/29/2016

56

Desired Skills

• Must have an interest in data visualization

• Think creatively

Required

• Interpersonal Skills

• In good academic standing

• Open to learning new skills

Technical Skills

• Problem Solving

• Programming skills (know at least one programming language)

BYRD VISUALIZATION LABUNDERGRADUATES

Participate in Local and National Visualization Challenges

Other Opportunities

• Summer Research Experiences

• Travel to Professional Conferences

• Conference Presentations

• Professional Development

• Challenging and rewarding work

Take CGT 270 Spring 2017

HOW TO GET INTO THE BYRD VIS LAB

9/29/2016

57

Spring 2017• Take an in-depth look

at data– When (Temporal Data)

– Where (Geospatial Data)

– What (Topical Data)

– With whom (Trees and Networks)

Where: Geospatial DataImage Source: Hexagon Geospatial

When: Temporal DataImage Source http://www.slideshare.net/kristw/finding‐patterns‐in‐temporal‐data

What: Topical DataImage Source http://blog.hubspot.com/agency/science‐brains‐crave‐infographicsWith Whom: Trees and Networks

Image Source http://blog.hubspot.com/agency/science‐brains‐crave‐infographics

CGT 581

• Spring 2016

• Fall 2016

9/29/2016

58

Stop by and see me!

KNOY 371Office Hours: Monday’s 10 AM – 11 AMORBy Appointment

Vetria L. Byrd

Assistant Professor

Computer Graphics Technology

[email protected]

Purdue Polytechnic Institute

https://polytechnic.purdue.edu/profile/vbyrdhttp://web.ics.purdue.edu/~vbyrd/@VByrdPhD, @BPViz, @VisREU

Thank You Image Source:http://careerconfidential.com/category/thank‐you‐notes/http://careerconfidential.com/wp‐content/uploads/2015/02/ThankYou2.jpg

introduction to data visualization - purdue...

Documents