SAS/SPECTRAVIEW Software User's Guide |
To
use SAS/SPECTRAVIEW, your data must
be stored in a SAS data set, which consists of variables and observations.
A variable is a column in the data set, such as quantities
or characteristics being measured, that has attributes such as a name and
a type (character or numeric). An observation is the
horizontal component of the data set, such as collections of values associated
with a single entity; each observation contains one value for each variable
in the data set.
The SAS data set that you use must have at least four
variables:
-
three variables,
which can be character or numeric, to be the axis variables,
that is, the X variable, Y variable, and Z variable. Each axis variable must
contain at least two unique values. For best results, the axis variables should
contain discrete data, which consists of distinct values
(noncontinuous) containing natural gaps, like patient IDs and years.
-
a
fourth variable, which must be numeric, to be the response variable. The response variable must contain at least two unique values.
For best results, the response variable should contain measured or modeled
response values (like sales or population) that are related in some way to
the axis values.
You can specify an optional fifth variable, which can
be character or numeric, as a BY variable. A BY variable allows you to
animate an image so that you can see how response values change according
to some grouping, like over time.
|
How the Software Displays Data |
When
you load a SAS data set into SAS/SPECTRAVIEW,
the software does the following:
-
The software reads each observation from the data
set and creates a three-dimensional volume grid by plotting
the values for the axis variables, which creates the x,y,z coordinates. That
is, for each observation, the following occurs:
- The value for the X variable is plotted on the
X axis, which is the horizontal axis.
- The value for the Y variable is plotted on the
Y axis, which is the vertical axis.
- The value for the Z variable is plotted on the
Z axis, which is the depth axis.
The
volume grid is actually an invisible network of lines that intersect in three-dimensional
space. Each intersection of an x,y,z coordinate is a location in three-dimensional
space, referred to as a data point. The shape and size of the volume grid
is determined by the number of unique X, Y, and Z values and is displayed
with a bounding box, which is a set of lines that outline
the three-dimensional volume grid.
- The software divides the values for the response
variable (the response values) into ranges, then color codes them using default
ranges and preset colors. According to the response value (and its associated
response value range color), the software then maps an appropriate color to
each data point that has an associated response value. Note that you
can customize the ranges and colors as explained in Setting Response Value Colors for Images.
Spatial Data
The
variables that you specify for the axes frequently (but not always) represent
dimensions of spatial data. For example, in a spatial diagram like the following
cubic volume, the x,y,z coordinate 6,5,5 represents a location that is 6 ticks
along the X axis, 5 ticks along the Y axis, and 5 ticks along the Z axis (counting
from an origination point shown as 0 in this figure):
Coordinate 6,5,5 in Three-Dimensional Space
To illustrate the relationship among the axis
values
and the response values, consider the following spatial data example:
To determine the age of someone sitting in a specific
seat in a stadium, you need to know the section, the row, and the seat number
to locate that person. The three values are x,y,z coordinates that identify
a specific location in space, which in this case is a stadium. Once located,
the person can be asked his age; that number becomes the fourth value...the
response value. You could collect the same information for everyone in the
stadium. That is, you could attach an age response value to each location
identified by section, row, and seat.
If you created a data set of the seating information
and loaded it into SAS/SPECTRAVIEW,
the following would occur:
- The resulting data points would form a volume
grid displayed within a bounding box.
- The data points would represent the seat holders.
- The response value for each
data point would be
the age of the occupant.
You could then explore the data visually and determine,
for example, whether age groupings occur in various locations in the stadium.
You could also display any empty seats, which have no response value at that
location.
Locations in three-dimensional space are similar to
stadium seat locations. For example, if you want to test the amount of sulphur
in the air at various locations, you would need three coordinates similar
to section, row, and seat. These might be 20 km east, 10 km north, and 200
meters up. The coordinates describe a specific location in space where a
sulphur sample can be taken and recorded. When you display the data, a color
is mapped to each response value, representing ranges of values, for example,
values between 0.0 and 0.5 could be red, values between 0.51 and 1.0 could
be yellow, and so on.
Non-Spatial Data
Since
the axis variables represent different dimensions of data, you can use SAS/SPECTRAVIEW to
explore non-spatial data as well.
For example, the sample data set MORTGAGE (which contains
mortgage payments for various numbers of years, interest rates, and loan amounts)
can be represented several ways. The axis variables could be principal amount,
percentage rate, and term of loan; it does not matter which variable you assign
to X, which to Y, and which to Z. This presumes that the PAYMENT variable
is the response you want to explore. Or you could assign the variables so
that AMOUNT is the response to explore, with axes of term, rate, and affordable
payment range.
Assume the following variables for SAS/SPECTRAVIEW:
RATE |
as the X variable, which is the loan
interest percentage rate. |
AMOUNT |
as the Y variable, which is the loan
amount. |
YEARS |
as the Z variable, which is the number
of years for the loan. |
PAYMENT |
as the response variable, which is
the monthly payment amount. |
SAS/SPECTRAVIEW reads
the data and generates the horizontal X axis (representing the loan interest
rate), the vertical Y axis (representing the loan amounts), and the depth
Z axis (representing the number of years for the loan). Each resulting data
point is applied a color representing a response value, which is a monthly
payment amount.
MORTGAGE Data Set Displayed as a Point Cloud
examines the data using a point cloud (which is one of the SAS/SPECTRAVIEW visualization
techniques), showing the relationship of the monthly payments to percentage
rate, loan amount, and length of the loan. This point cloud displays a subset
of the data points, showing only the higher loan payments (response values).
With this point cloud, you can determine that most of the higher loan payments
are for a shorter number of years.
MORTGAGE Data Set Displayed as a Point Cloud
|
Summary of Software Tools |
SAS/SPECTRAVIEW provides
the following tools that you use to create and analyze images.
Visualization Techniques
Visualization techniques create images representing your data.
-
-
Cutting planes produce slices
of data, either perpendicular to an axis or not perpendicular to any axis.
In addition to the axis cutting planes, you can request two- and three-dimensional
surface views (surfaces, charts, stacks, and plots) at a specific cutting
plane's
location.
-
-
Direct volume rendering creates a two-dimensional
image of the entire volume of data points with
transparency.
-
-
Isosurface produces a three-dimensional surface
by connecting all the data points with one response
value.
-
-
Point
cloud displays response values with colored
markers, showing individual data
points.
-
-
Solid-volume image produces a colored, three-dimensional
solid-block image of the data points, providing a surface view of the data
at the volume's borders.
Customization Options
SAS/SPECTRAVIEW provides
a variety of options that you can use with visualization techniques to aid
in data exploration and analysis:
-
-
Probe analysis displays the response value
for a specific data point in an
image.
-
-
Bounding box displays a set of lines that
outline the three-dimensional volume representing the data in the Volume window.
By default, the bounding box is on, but you can turn it
off.
-
-
Image annotation allows
you to add the following to two- and three-dimensional
images: axis labels that name the variables associated with the axes, major
and minor tick marks along the axes, text, and a response legend (a visual
key to the colored data
points).
-
-
Image transformations let you view a three-dimensional
image from any angle or in a different size. That is, you can rotate, move,
and zoom an
image.
-
-
Color palette lets you
customize colors for the response value ranges,
missing values, an isosurface, and image annotations (such as tick marks,
axes labels, text, and the bounding box). You can also save and recall user-defined
color palette files.
Processing Options
These options let you modify
your view
of the data:
- BY variable
-
is a variable specification whose values define groups of observations,
such as hour, month, or year. Specifying a BY variable allows you to animate
an image so that you can see how response values change according to some
grouping, like over time.
- Categorizing data
-
groups
numeric data to create discrete ranges for the X, Y, and Z axes, which results
in reducing the number of data points created for the volume grid.
- WHERE clause
-
allows you to specify a subset of data to be read into
the software.
- Data filtering
-
smooths
or sharpens data to deemphasize or highlight variations in response values.
The software provides several predefined filters, or you can create your own.
- Data saving
-
lets
you save all or part of the original data values to a SAS data set.
- Image saving
-
lets
you save a displayed image as a TIFF (Tagged Image File Format) file or a
PostScript file.
Copyright 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.