exploring high d spaces with multiform matrices and small multiples
Download
Skip this Video
Download Presentation
“Exploring High-D Spaces with Multiform Matrices and Small Multiples”

Loading in 2 Seconds...

play fullscreen
1 / 22

“Exploring High-D Spaces with Multiform Matrices and Small Multiples” - PowerPoint PPT Presentation


  • 134 Views
  • Uploaded on

“Exploring High-D Spaces with Multiform Matrices and Small Multiples”. MacEachren, A., Dai, X., Hardisty, F., Guo, D., and Lengerich, G. Proc. IEEE Symposium on Information Visualization (2003), 31–38. http://www.geovista.psu.edu/. Mudit Agrawal Nathaniel Ayewah. The Plan. Motivation

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about ' “Exploring High-D Spaces with Multiform Matrices and Small Multiples”' - bianca-atkins


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
exploring high d spaces with multiform matrices and small multiples

“Exploring High-D Spaces with Multiform Matrices and Small Multiples”

MacEachren, A., Dai, X., Hardisty, F., Guo, D., and Lengerich, G.

Proc. IEEE Symposium on Information Visualization (2003), 31–38.

http://www.geovista.psu.edu/

Mudit Agrawal

Nathaniel Ayewah

the plan
The Plan
  • Motivation
  • Contribution
  • Analysis Methods
  • GeoVISTA studio
  • Conclusions
motivation
Motivation
  • Discover Multivariate relationships
  • Examine data from multiple perspectives

DATA  INFORMATION

contribution
Contribution
  • Visual analysis of multivariate data
    • Combinations of scatterplots, bivariate maps and space-filling displays
    • Conditional Entropyto identify interesting variables from a data-set, and to order the variables to show more information
    • Dynamic query/filtering called Conditioning
contribution1
Contribution
  • Back-end: Design Box

Building of applications using visual programming tools

  • Front-end: GUI Box

Visualizing data using the developed designs

Source: GeoVista Studio

sorting
Sorting

Analysis Methods

  • Nested sorting – sort a table on selected attributes
  • To understand the relationships between sorted variables and the rest
  • Permutation Matrix :
    • cell values are replaced by graphical depiction of value.
    • Rows/cols can be sorted to search for related entities
    • e.g.
sorting1
Sorting

Analysis Methods

  • Augmented seriation:
    • Organizing a set of objects along a single dimension using multimodal multimedia
  • Correlation matrices
  • Reorderable Matrices:
    • Simple interactive visualization artifact for tabular data

Source: (Siirtola, 1999)

space filling visualization
Space-filling visualization

Analysis Methods

Mosaic plot

Sunburst methods

Source: (Schedl, 2006)

Source: (Young, 1999)

Pixel-oriented methods

Source: (Keim, 1996)

multiform bivariate small multiple
Multiform Bivariate Small Multiple

Analysis Methods

  • Small Multiples

A set of juxtaposed data representations that together support understanding of multivariate information

Source: (MacEachren, 2003)

multiform bivariate matrix
Multiform Bivariate Matrix

Analysis Methods

Source: (MacEachren, 2003)

demonstration
Demonstration
  • Basic Demo
    • Application construction
    • Scatterplot, Geomap
    • Dynamic linking, eccentric labeling etc.
high dimensionality
High Dimensionality
  • Interactive Feature Selection
    • Guo, D., 2003. Coordinating Computational and Visualization Approaches for Interactive Feature Selection and Mulivariate Clustering. Information Visualization 2(4): 232-246.
high dimensionality1
High Dimensionality
  • “Goodness of Clustering”
    • high coverage
    • high density
    • high dependence
  • E.g.
    • Correlation
    • Chi-squared
    • Conditional Entropy

HIGH

HIGH

LOW

conditional entropy

1

2

3

4

1

2

Conditional Entropy
  • Discretize two dimensions into intervals
    • Nested Means

mean

mean

mean

Source: (Guo, 2003)

conditional entropy1
Conditional Entropy

Source: (Guo, 2003)

ordering dimensions
Ordering Dimensions
  • Related dimensions should be close together

Sort Method: Minimum Spanning Tree

Sort By: Conditional Entropy

A

5

B

9

15

16

21

C

D

4

Ordering: B A D C

unsorted

demonstration1
Demonstration
  • Advanced Demo
    • Interactive Feature Selection
    • PCP, SOM, Matrix
    • Conditioning
conclusions
Conclusions
  • Strengths
    • Dynamic Linking of different representations
    • Visualizing clusters of dimensions
    • Rich and extensible toolbox
  • Weaknesses
    • Usability
    • Arrangement of Windows
references
References
  • Guo, D., (2003). Coordinating Computational and Visualization Approaches for Interactive Feature Selection and Mulivariate Clustering. Information Visualization 2(4): 232-246.
  • Keim, D (1996) Pixel-oriented Visualization Techniques for Exploring Very Large Databases, Journal of Computational and Graphical Statistics.
  • Schedl, M (2006), CoMIRVA: Collection of Music Information Retrieval and Visualization Applications. Website. http://www.cp.jku.at/people/schedl/Research/Development/CoMIRVA/webpage/CoMIRVA.html
  • Siirtola, H. (1999), Interaction with the Reorderable Matrix. In E. Banissi, F. Khosrowshahi, M. Sarfraz, E. Tatham, and A. Ursyn, editors, Information Visualization IV \'99, pages 272-277. Proceedings International Conference on Information Visualization.
  • Young, F (1999), Frequency Distribution Graphs (Visualizations) for Category Variables, unpublished. http://forrest.psych.unc.edu/research/vista-frames/help/lecturenotes/lecture02/repvis4a.html.
ad