Databases Reference
In-Depth Information
Research suggests, however, that there are other issues to consider. Wolfe
found that as an image is presented, the image is immediately abstracted, but
details of the image are not retained in memory when focus shifts to a different
image. According to Healey, “Wolfe's conclusion was that sustained attention
to the objects tested in his experiments did not make visual search more
efficient. In this scenario, methods that draw attention to areas of potential
interest within a display [i.e., preattentive methods] would be critical
in
allowing viewers to rapidly and accurately explore their data”.
Based on this research, VisMiner was created to present multiple, concurrent,
non-overlapped visualizations designed specifically to be preattentively proc-
essed. The preattentive properties allow you to immediately recognize patterns
in the data and the multiple, concurrent views supplement working memory as
your eyes move from visualization to visualization when comparing different
views of the dataset. As VisMiner was designed, information density was not
considered as important when choosing the types of visualizations to be
incorporated. Since information-dense visualizations require a relatively longer
“study” time to extract patterns,
they were not considered to be viable
extensions to working memory.
Tutorial - Using VisMiner
Initializing VisMiner
If you have not already done so, start the VisMiner Control Center on one of the
computers that you will be using for your data mining activities. Upon start-up,
the Control Center will open in a maximized window shown in Figure 2.3. The
Control Center is divided into three panes:
Available Displays - The upper left pane depicts available displays that
have been connected to the Control Center via VisSlave. Initially this pane is
blank, because no slaves are currently connected.
Datasets and Models - The lower left pane is used to represent open
datasets, derived datasets, and models built using the open or derived
datasets. Again, upon start-up, this pane is blank.
Modelers - The right side pane contains icons representing the VisMiner
data mining algorithms (or modelers) available for processing the data-
sets. Each is represented by a gear, and overlaid by the title of the modeler.
The Control Center interface is designed for direct manipulation, meaning
that you perform data mining operations by clicking on or dragging and
dropping the objects presented. Once you get started, you will find the use
of VisMiner very intuitive and easy to learn.
