Databases Reference
In-Depth Information
3.2.1 BACKGROUND
There are several fields that are core to visual data mining, including the following: information
visualization, knowledge discovery, and data mining. Information visualization has been introduced
in the previous subsection. In the sequel, we introduce knowledge discovery and data mining.
Knowledge Discovery
Knowledge discovery (KD) may be defined as the process of identifying valid, novel, potentially
useful, and ultimately understandable models and/or patterns in data (
Fayyad et al.
,
1996a
,
b
). On
the whole, the knowledge discovery process may be defined as an interactive and iterative non-trivial
process that entails various phases as seen in Figure
3.19
.
Interpretation /
Evaluation
Data Mining
Transformation
Knowledge
Preprocessing
Selection
Patterns
Transformed
Data
Preprocessed Data
Data
Target Data
Figure 3.19:
The Knowledge Discovery process (based on
Fayyad et al.
,
1996a
).
The KD phases include the following: carrying out some initial planning (understanding the
application domain, relevant prior knowledge, and goal(s) of the user), data integration, selection
of target data, data cleaning and pre-processing, data reduction and transformation, selection of
suitable data mining techniques to support the discovery process, and evaluation, presentation and
interpretation of results. Through carrying out the phases, the KD process intends to find a subset
of results that may be considered as new “knowledge” (
Fayyad et al.
,
1996a
,
b
). KD is of interest to
researchers in many research disciplines such as the following: machine learning, pattern recognition,
databases, statistics, artificial intelligence, expert systems, and information visualization.
Data Mining
Data mining is a step in the knowledge discovery process that, under acceptable computational
efficiency limitations, enumerates models and patterns over the data (
Fayyad et al.
,
1996a
,
b
). Data
mining methods include the following: clustering, classification, regression, characterization, depen-
dency modeling change and deviation detection, and pattern-based similarity matching.
