Databases Reference
In-Depth Information
The primary goals of data mining are verification and discovery. The verification goal aims at
validating some hypotheses based on specific user needs. The user generates a series of hypothetical
patterns and relationships (assumptions or claims). The user then formulates and issues queries to the
system (actually, to the data itself ), and to verify (or disprove) the claims. The discovery goal involves
finding “new” patterns or discovering new knowledge. Rather than verify hypothetical patterns, the
goal here is to use the data to uncover or identify such patterns. Such methods of discovery may
be initiated based on the guidance of a user to analyze a certain domain through a predetermined
perspective or by automated learning. Discovery can be predictive or descriptive. Prediction entails
“foretelling” unknown or future values of the same variables or other variables of interest whereas
description involves getting an interpretation/understanding of the data. Classification, regression
and time series models are primarily useful for prediction. Clustering, association and sequence
discovery models are primarily useful for description of the behavior that is captured in the data.
3.2.2 DEFINITION AND IMPORTANCE OF VISUAL DATA MINING
In the KD or data mining process, only the user can ultimately or directly determine the usefulness or
value of some resulting knowledge in a specific domain or application. On the same note, different
users may rate the same knowledge very differently. What one user considers valuable may be
deemed to be of less or no value by another user (
Ankerst, M.
,
2001
). The human user is potentially
resourceful and instrumental in guiding or steering the entire discovery process. Human involvement
in the KD process is pivotal in the mining and acquisition of useful knowledge. Moreover, the
human-vision channel has outstanding capabilities that enable both recognition and understanding
of overwhelming data in an instant (
Card et al.
,
1999
). Tapping into that channel would primarily
entail appropriately exploiting visual strategies within the user interface.
The foregoing discussion points us to the field of
visual data mining
. There exist various
definitions of visual data mining in the literature. “Visual data mining is the use of visualization
techniques to allow data miners and analysts to evaluate, monitor, and guide the inputs, products
and process of data mining” (
Ganesh et al.
,
1996
).
Ankerst, M.
(
2000
) defines visual data mining
as “a step in the KD process that utilizes visualization as a communication channel between the
computer and the user to produce novel and interpretable patterns.”
Kopanakis and Theodoulidis
(
2001
) say that visual data mining “involves the invention of visual representations during all three
data-mining life cycle stages, as partitioned to the data preparation, model derivation and validation
stage.” According to
Simoff, S.
(
2001
): “Visual data mining (
Michalski et al.
,
1999
) is an approach
to explorative data analysis and knowledge discovery that is built on the extensive use of visual
computing (
Gross, M.
,
1994
;
Nielson et al.
,
1997
). The basic assumption is that large and normally
incomprehensible amounts of data can be reduced to a form that can be understood and interpreted
by a human through the use of visualization techniques based on a particular metaphor or a com-
bination of several metaphors (preferably, but not necessarily preserving the consistency of their
combination).”
