JViz User's Manual


Getting Up and Running

SYSTEM REQUIREMENTS: 
	Java 1.1.2 or higher required.  Java 1.1.5 or higher recommended
	Java Foundation Classes (aka Swing) version 1.0 or higher
		This is a package from Sun Microsystems and can be downloaded
		from their web site at: http://www.javasoft.com/jfc

INSTALLATION:
	Once the JViz files are unarchived, the included scripts require two
	environment varibles be set.
		JAVA_HOME - This environment variable should point to the top
			    directory of the installed JDK or JRE.
		SWING_HOME - This environment variable should point to the top
			     directory of the Swing package installation


Overview

JViz allows users to visually explore multivariate data in a variety of
methods.  The major concepts which users need to grasp to be able to
effectively utilize JViz are the data projection techniques available and
the use of the N-dimensional brush.  These concepts are outlined below,
after which a description of the JViz interface is presented. 

JViz display methods

JViz supports three ways of visualizing data: scatter plots, glyphs, and
parallel coordinates.  Each of these displays can be selected by pressing
the corresponding button on the JViz main display. 

Scatter plots

Scatter plots are one of the oldest and most commonly used methods to
project high-dimensional data to 2-dimensions.  In this method, a grid of
parallel projections of the data is generated.  Each of these is a simple
plot for the two dimensions it represents.  The horizontal dimension of a
scatter plot is controlled by the column it resides in, and likewise the
vertical dimension is controlled by the row.  At the top of each column is
a numbered label and at the left of each row is the number that shows what
dimensions that row / column represents. 

Glyphs

The definition of a glyph covers a large number of techniques which map
data values to various geometric and color attributes of graphical
primitives or symbols.  Histograms are perhaps the most widely recognized
form of glyph, where data values control the height of rectangular bars. 
One might even consider a scatter plot a specialized form of glyph
representation, where data values control the positional attributes of a
symbol.  In JViz, we use the star glyph pattern, which creates for each
data point N rays emanating at equal angles from a point on the screen.
Currently glyphs are spaced uniformly across the screen, though we might
want to use 2 of the dimensions to control position.  The length of each
ray is determined by the data value for that dimension.  A polyline is
then generated to encompass the rays, forming a blob shape.  The glyph
brush tool may be used as a key to the dimensions. 

Parallel Coordinates

In Parallel Coordinates, each dimension corresponds to an axis, and the N
axes are organized as uniformly spaced vertical lines.  A data element in
N-dimensional space manifests itself as a connected set of points, one on
each axis.  Points lying on a common line or plane create readily
perceived structures in the image.  In generating the display of parallel
coordinates in JViz, the view area is divided into N vertical slices of
equal width.  At the center of each slice an axis is drawn, along with a
label at the top end.  Data points are generated as polylines across the N
axes. 


Brushing

A big problem with projecting data of a given number of dimensions to a
smaller number of dimensions is that you invariably lose information
regarding the spatial relationships of the data.  N-D brushing is a way of
recovering some of this lost information by highlighting data points that
fall into a user-defined subspace.  Thus, one could say ``show me the
points within a (Euclidean) distance of N from a certain location.'' A
brush is completely defined by its shape, size, and location.  We assume a
hyper-parallelepiped (a fancy name for an N-D box), and the user specifies
the size of the box in each dimension.  Then a location in N-D space is
specified as a center point for the brush.  Any data point falling within
the N-D brush can have an operator applied to it, such as highlighting or
masking. 

Brush Manipulation

Each of the display methods allows the user to directly manipulate the
brush.  A summary of manipulation techniques for each display method
follows: 
  
Scatter plots -- The brush coverage on the scatter plot display is
represented by a series of rectangular boxes, one for each scatter plot.
The boxes represent the range of points that the brush covers for the two
dimensions of a particular scatter plot.  Each of these boxes may be
manipulated using the mouse.  Clicking the mouse button will reshape the
brush to the coordinates of the point of the cursor.  To recenter the
brush hold down both the Control key and mouse button and move the whole
brush. 

Glyphs -- The brush coverage is not displayed in the glyph view.  The
brush may be recentered by clicking on a glyph with the left mouse button.
The brush will change location to be centered on the selected point.  This
is useful for finding glyphs that are similar in shape to an existing
glyph. 

Parallel Coordinates -- The brush coverage on the parallel coordinate
display is represented by a shaded region that spans all the dimensions.
The shaded region along an axis represents the range of points that the
brush covers along that axis.  Brush manipulation in this display works
similarly to the scatter plot display.  To change the value of the brush
in a single dimension, move the cursor to the axis and click the mouse
button.  If the value is above the center of the brush, then the maximum
brush value will change.  Otherwise, the minimum brush value will change. 
To reshape the brush to fit selected data, hold down both the Shift key
and the mouse button while moving the cursor over the desired data. 
Moving the entire brush is the same in parallel coordinates as it is in
scatter plots.


Brush Manipulation Tools

This is an interface to the Glyph Brush Tool.  It is located in the
'Display' menu under the heading "Glyph Brush Tool". 


Glyph Brush Tool -- This tool represents an enlarged glyph.  The currently
active brush coverage is displayed as a shaded region on the glyph.  The
brush may be resized by clicking on the appropriate ray of the glyph.
These manipulations affect a local copy of the brush only, in order to
have these changes take effect on the true brush, you must click the Apply
button. 

                      

Painting

Painting is a method of creating a brush from the presented data.  This
method is currently available on in parallel coordinates, but future
versions of JViz plan to include painting with scatter plots.  Painting is
accomplished by holding down the Shift key on the keyboard and dragging
the mouse over points of interest.  As long as the Shift key is held down
the mouse cursor will continue to behave as a virtual paintbrush.  When
the Shift key is released, a brush will be generated that contains all the
points that were painted. 


Visualization Hints

The following are hints that can be used to find patterns in data in each
of the view methods. 

Scatter plots -- look for linear features, which indicate a pairwise
relationship within the data.  Clusters are a bit deceptive, as there may
be quite a spread amongst the other dimensions. 

Glyphs -- look for similar shapes and simple transitions between shapes,
e.g.  a convexity that grows or shrinks.  It is also fairly easy to spot
anomalies, although if there are too many shapes the anomalous ones may
not stand out.  One trick for examining dense data with 2 spatial
dimensions is to use square data sets (e.g. width = height).  Because the
grid of glyphs uses the square root of the data set size as the number of
rows and columns, you in effect get 2 dimensions for free.  Regions of
similarity and boundaries between distinct regions then manifest
themselves as textures.

Parallel Coordinates -- relationships between adjacent dimensions manifest
themselves as line segments with similar orientation.  Adjacent dimensions
with negative correlation create an `X' pattern between them. 

Input File Format

The file format for the data sets read by JViz is in simple ASCII format. 
All dimensions are represented by floating point values; discrete integer
values will be converted automatically, but nominal values must be mapped
to floating point.  Simply assigning an integer to each name in the
dimension usually works fine.  The data file is composed of four main
sections, described in detail here: 
  
 
Section 1 -- The first line in the data set file specifies how many
dimensions there are in the data and how many data points.  These values
are stored as ASCII strings and must be separated by whitespace. 
 
Section 2 -- The next section of the input file contains the labels for
each of the dimensions.  These are stored as ASCII strings and there must
be one name per line.  The number of lines in this section of the file
must match with the number of dimensions specified on the first line. 
 
Section 3 -- The third section of the input file contains the minimum and
maximum values for each dimension and the number of bins to use in
dimensional stacking.  These are written as floating point numbers in
ASCII format on a single line.  These are stored as min/max/bin triplets
for each dimension as such:
  
min1 max1 bin1 min2 max2 bin2 ... 
 
The number of min/max/bin triplets should match the number of dimensions
specified on line 1. 
 
Section 4 -- The final section of the input file contains the actual data
points.  Each sample is stored as a set of ASCII format floating point
numbers on a single line.  The number of floating point values on a line
must match the number of dimensions specified on line 1, and the number of
samples should match the number also specified on line 1. 

See JViz/data for some examples of JViz data files.