Scatterplots remain one of the most popular and widely-used visual representations for multidimensional data due to their simplicity, familiarity and visual clarity, even if they lack some of the flexibility and visual expressiveness of newer multidimensional visualization techniques. This paper presents new interactive methods to explore multidimensional data using scatterplots. This exploration is performed using a matrix of scatterplots that gives an overview of the possible configurations, thumbnails of the scatterplots, and support for interactive navigation in the multidimensional space. Transitions between scatterplots are performed as animated rotations in 3D space, somewhat akin to rolling dice. Users can iteratively build queries using bounding volumes in the dataset, sculpting the query from different viewpoints to become more and more refined. Furthermore, the dimensions in the navigation space can be reordered, manually or automatically, to highlight salient correlations and differences among them. An example scenario presents the interaction techniques supporting smooth and effortless visual exploration of multidimensional datasets.

1. “ ”
Rolling the Dice: Multidimensional Visual
Exploration using Scatterplot Matrix Navigation
Niklas Elmqvist | Purdue University
Pierre Dragicevic | INRIA
1
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation
Jean-Daniel Fekete | INRIA

2. Visualizing Complex Data
Complex visualization
?
Complex dataset
Many simple visualizations
2
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

3. Motivation
 Information visualization is now more than 15
years old
 Lots of visual representations
 New ones every year at IEEE InfoVis
 Varying complexity
 Varying purpose
 Scatterplots: one of the simplest and most
widely used visual representations
 Multidimensional data
 Examples: SpotFire, ADVIZOR, etc
3
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

4. Scatterplots
dimension B
 Assign dimensions to
graphical axes
 Two (typical) or three
 Data cases as points in the
space defined by axes
 Additional dimensions
 Point color
 Point size
 Point shape
…

 Limited number of
dimension A
displayed dimensions!
4
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

5. Problem
 Many (most?) interesting real-world datasets
have many dimensions
 Certainly more than a scatterplot can show!
 Two possible solutions:
1. More complex visual representation
Parallel coordinates, DataMeadow, etc…

 Lose simplicity of scatterplots
2. Create many scatterplots (one per combination of
dimensions)
 But how to visualize them?
5
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

6. Scatterplot Matrices
 Idea: Create matrix of all data dimensions
 Each column/row is a data dimension
 Each cell in the matrix is a scatter plot
 Reordering of columns/rows (automatic or manual)
 Explored by others, but we add a twist:
 Use matrix as a space for navigation
 Visual exploration becomes a navigation problem
 Result: Visualize complex data through
sequence of simple visualizations
6
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

7. Example: Scatterplot Matrix
7
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

8. Navigating Scatterplots
 Idea: Navigating from scatterplot A to scatterplot
B using an animated transition
 Problem with linear interpolation animation
 No semantic meaning to the user
 Can be complex to follow
 Compounded by large datasets
 Can we add meaning to the transition?
8
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

9. Adding Meaning to Transitions
 Make the transition between scatterplots in the
scatterplot matrix easier to follow:
 Restrict to rectilinear movement (no diagonal
movement)
 Change only one visualized data dimension at a time
 Utilize unused third graphical dimension for the new
dimension to show during the transition
 Metaphor: rotating a 3D scatterplot to show new 2D
projection
 Like rolling dice to see another side
9
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

10. Using the Third Dimension
 Lazy allocation 3rd graphical dimension
 Invisible when viewing a 2D scatterplot, so can be
assigned any data dimension
 Allocation performed only during the transition
 Not “true” 3D – we use third dimension as a
presentation tool only
 Mechanics of performing 3D rotation:
 Perspective view: 3-stage animation
(extrusion, rotation, projection)
 Looks more natural, but requires three steps
 Orthographic view: 1-stage animation (rotation)
 Only one step, but may look “weird”
10
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

11. Example: 3D Perspective Transition
11
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

12. Putting It Together
 ScatterDice application
 Built in Java and uses OpenGL through JOGL
 Loads multidimensional datasets using the InfoVis
Toolkit [Fekete 2004]
 Exploration by navigating a scatterplot matrix
 3D transitions between adjacent scatter plots
 Support reordering of dimensions
 Automatic: by correlation between dimensions
 Manual: drag and drop of rows columns
 Interaction for the navigation is important
12
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

13. Interaction Techniques
13
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

14. Demonstration
14
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

15. Conclusions and Future Work
 Idea: Combine multiple simple visualizations
using interactive animation
 Visualize complex datasets
 Provide meaningful transitions
 Interaction is a key element
 Avoid automated tours
 Allow refinement of queries
 Future work
 Empirical evaluation of this method
 Other applications of the same approach?
15
Rolling the Dice: Multidimensional Visual Exploration using Scatterplot Matrix Navigation

16. Questions?
 Niklas Elmqvist
Electrical & Computer Engineering
Purdue University
West Lafayette, IN 47907-2035
E-mail: elm@purdue.edu
Web: http://engineering.purdue.edu/~elm/