Database Reference
In-Depth Information
mining methods. In particular, frequent patterns can be used to determine the key
segments in the trajectories which are used frequently over time.
Image and Multimedia Data Mining:
In this case, features of images can be
treated as attributes in transactions, and frequent patterns can be determined from
these transactions in order to determine the important characteristics of images.
Such characteristics can be used for a variety of mining tasks. Image data are
closely related to spatial data, since the pixels in an image have spatial attributes
as well as non-spatial attributes.
Chemical and Biological Applications:
Frequent patterns can be used to de-
termine important motifs in a variety of chemical and biological applications.
In many cases, these correspond to frequent patterns in graphs and structured
data. Examples include toxicological analysis, chemical compound prediction,
phylogenetic and RNA analysis.
This chapter will provide an overview of the afore-mentioned applications of frequent
pattern mining. The number of possible applications of frequent pattern mining are
varied, and arise in many domains. For example, different kinds of applications are
possible within the context of set-based data (e.g. market baskets), graph-based data,
or graphs represented as trees. While this chapter provides an idea of the landscape,
the main goal is to cover the key scenarios in which frequent pattern mining can
be applied. This will provide the reader the machinery for understanding how these
techniques can be useful in different contexts.
This chapter is organized as follows. Customer analysis applications are discussed
in Sect.
2
. In Sect.
3
, we discuss the problem of using frequent patterns for clustering.
The problem of using frequent pattern mining for classification is discussed in Sect.
4
.
Applications of frequent pattern mining to outlier analysis are discussed in Sect.
5
.
Methods for using frequent pattern mining methods in indexing are discussed in
Sect.
6
. The use of frequent pattern mining methods in Web-related mining tasks is
discussed in Sect.
7
. Text applications of frequent pattern mining are discussed in
Sect.
8
. Applications for temporal data are discussed in Sect.
9
. Methods for using
frequent pattern mining for analyzing spatial and spatio-temporal data are discussed
in Sect.
10
. Methods for software bug detection are discussed in Sect.
11
. Methods for
mining biological and chemical data are discussed in Sect.
12
. Section
13
discusses
resources for the practitioner, which includes the key commercial and open-source
software available for frequent pattern mining. The conclusions and summary are
discussed in Sect.
14
.
2
Frequent Patterns for Customer Analysis
The motivating application for frequent pattern mining was proposed in the context
of supermarket and customer analysis [
13
]. In this case customer behavior is captured
either by baskets of items bought together or by sequences of items which are bought
in succession. Frequent patterns can be used in order to determine the common
patterns of buying behavior. A rather old, but much used example of a frequent pattern
