Graphics Reference
In-Depth Information
Introduction
7.1
In the context of data visualization, a glyph is a visual representation of a piece of
data where the attributes of a graphical entity are dictated by one or more attributes
of a data record. For example, the width and height of a box could be determined by
a student's score on the midterm and final exam for a course, while the box's color
mightindicatethegenderofthestudent.hedefinitionaboveisratherbroad,asitcan
cover such visual elements as the markers in a scatterplot, the bars of a histogram,
or even an entire line plot. However, a narrower definition would not be su
cient
to capture the wide range of data visualization techniques that have been developed
over the centuries that are termed glyphs.
Glyphs are one class of visualization techniques used for multivariate data. heir
major strength, as compared to techniques such as parallel coordinates, scatterplot
matrices, and stacked univariate plots, is that patterns involving more than two or
three data dimensions can oten be more readily perceived. Subsets of dimensions
can form composite visual features that analysts can be trained to detect and classify,
leading to a richer description of interrecord and intrarecord relationships than can
be extracted using other techniques.
However,glyphsdohavetheirlimitations. heyaregenerally restrictedintermsof
howaccuratelythey can convey data duetotheir sizeand thelimits of ourvisual per-
ception system to measure different graphical attributes. here are also constraints
on the number of data records that can be effectively visualized with glyphs; exces-
sive data set size can result in significant occlusion or the need to reduce the size of
each glyph, both of which make the detection of patterns di
cult, if not impossi-
ble.husglyphsareprimarilysuitableforqualitativeanalysisofmodest-sizeddata
sets.
his paper describes the process of glyph generation - the mapping of data at-
tributes to graphical attributes - and presents some of the perceptual issues that can
differentiate effective from ineffective glyphs. Several important issues in the use of
glyphs for communicating information and facilitating analysis are also discussed,
includingdimension orderandglyphlayout.Finally, someideas forfuturedirections
for research on visualization using glyphs are presented.
Data
7.2
Glyphs are commonly used tovisualize multivariate data sets. Multivariate data,also
called multidimensional or n-dimensional data, consist of some number of items or
records, n, each of which is defined by a d-vector of values. Such data can be viewed
as a dxn matrix, where each row represents a data record and each column repre-
sents an observation, variable, or dimension. For the purpose of this paper, we will
assumea data itemis avector of scalar numeric values. Categorical andother nonnu-
meric values can also be visualized using glyphs, though oten only ater conversion
tonumeric form (Rosario et al.,
).Nonscalar values can also be incorporated by
