Table of Contents
Abstract
Introduction
Technologies
Data Warehouse
Examples
Characteristics
Platforms
Summary
|
|
4.0 Characteristics of Data Mining
Figure 2, shows the data and information flow when the data mining
techniques
described in Section 3.0 are used. The parallel streams indicate that
cost-effective, scalable parallelism is a critical technology for data
mining.
Data mining provides insights that are hard to come by using traditional
techniques. The traditional techniques described in Section 1.0 complement
the
new data mining techniques. They cannot discover the kinds of information
that
were described before.
For example, the technique of mining for association rules allows the
analyst
to discover rules linking items in different categories. The techniques
described in Section 1.0 work best at discovering rules within a certain
category. The sheer number of possible rules between items in different
categories is daunting, and only the new techniques have been successful
in
mastering this complexity.
Also, unlike techniques that mechanically aid an analyst in the discovery
task,
data mining techniques truly perform a kind of discovery. Typically, the
results of data mining help to focus the analyst's attention. This, in
turn,
promotes the discovery of additional opportunities that are ripe for
exploitation. However, discovery is hard, and data mining techniques may
need
to be adapted to the application domain.
Data mining analysis tends to be bottom-up, and the best techniques have
been
developed with an orientation towards large volumes of data. This is
important
in the context of the data warehouse, where a typical enterprise usually
wishes
to use as much of the collected data as possible to arrive at reliable
conclusions and decisions.
|
|
