Aspects of Data in Visualization: Overview 

Characterization of data

Data

• "data" = information that can be represented in computer readable format
• data model = conceptual view of data
• data model ­ data format (physical view of data)
• choose expressive/effective visualization technique
• avoid "mental road blocks"

[BRO92], [GAL94]

Overview of selected data characteristics

• nominal, ordinal, quantitative
• point, scalar, vector
• "continuous" data
• topology/structure for non-continuous data
• data reliability
• valid range of data 
• time descriptors

Nominal, ordinal, quantitative

• Nominal data
  • members of certain class, e.g. [Georgia, Florida, North Carolina, Delaware], or [Maple, Birch, Oak]
  •  effective visual attributes: color - hue, symbol
• Ordinal data
  • related by order, e.g. [low, medium, high], or [tiny, small, medium, large]
  • effective visual attributes: brightness, size, (color - hue)
• Quantitative data
  • carry precise numerical value, e.g. [2.3, 4.56, 0.8, 2.5E-35] 
  • effective visual attributes: position, length, (color - hue) 

Priorities of Visual Attribute for Various Data Types (Excerpt) [MAC86]

Point, Scalar, Vector

Syntactical categories, additionally characterized by dimensions

• Point
  • each data element is considered as a position in n-dimensional space.
  •  example: measurements of leaves: [length, width, tree type, age], e.g.
    [2.3, 1.2, B, 1], [4.3, 2.2, B, 3], [1.5, 1.5, M, 1], [3.0, 2.9, M, 3], ....
  •  expressive visualizations: scatter plots, glyphs
• Scalars
  • each data element has a numeric expression 
  • example: topography of terrain, expressed as 2-d field containing elevations 
• Scalar arrays
  • often "discrete samples of continuous functions"
  • usually 1 (linear), 2 (image) , or 3 (volumetric) dimensional data sets; samples in equidistant or non-equidistant steps. 
  • expressive visualizations: line graph, shaded surface, volume viewing
• Vectors
  • each data element is considered as a straight directed line with a certain length (magnitude) in n-dimensional space. 
  • example: Direction of particle flow in channel. 
  • expressive visualizations: arrows, stream lines, particle tracks

"Continuous" data

"Continuous" data can be represented by (samples of) function:

yi = fi (X), where X = (x1 , x2 , x3 , ..., xn ); i=[1,....,m]

 

x .... independent variables; e.g space, time, spectral ("dimensions")

 

y .... dependent variables ("parameters")

 

x, y large ... multidimensional, multiparameter, multivariate data

 

=> regular/irregular format

Expressive visualizations of functions: similar to scalar, quantitative, ordinal

Interpolation methods: must be meaningful in problem space

Computation time for visualization techniques faster on regular grids

Note: for valid data must avoid aliasing artifacts

Topology/structure of non-continuous data

• Types of topology/structure, e.g.
  • sequential (text) 
  • hierarchical
  • relational
  • single points and connectors

• Examples and corresponding expressive visualizations
  • molecules (ball-and-stick model)
  • data bases (cone tree; perspective wall) 

Other data characteristics

• Data reliability: Missing data or unreliable data
  • expressive visualizations: error bars; indicate borders between real/missing data
  • careful with interpolation
• Valid range of data: min / max / mean / median
•  Time descriptors
  • Various meanings of time: simulation time, simulated/actual time frame,computation time, recording and playback time, user's time frame
  • "time models" to support time conversions necessary to synchronize

Aspects of Data
HyperVis Table of Contents

Last modified on February 11, 1999, G. Scott Owen, owen@siggraph.org