Data Warehousing for Association Mining

IntroductIon With the phenomenal growth of electronic data and information, there are many demands for developments of efficient and effective systems (tools) to address the issue of performing data mining tasks on data warehouses or multidimensional databases. Association rules describe associations between itemsets (i.e., sets of data items) (or granules). Association mining (or called association rule mining) finds interesting or useful association rules in databases, which is the crucial technique for the development of data mining. Association mining can be used in many application areas, for example, the discovery of associations between customers' locations and shopping behaviours in market basket analysis. Association mining includes two phases. The first phase is called pattern mining that is the discovery of frequent patterns. The second phase is called rule generation that is the discovery of the interesting and useful association rules in the discovered patterns. The first phase, however, often takes a long time to find all frequent patterns that also include much noise as well (Pei and Han, 2002). The second phase is also a time consuming activity (Han and Kamber, 2000) and can generate many redundant rules (Zaki, 2004) (Xu and Li, 2007). To reduce search spaces, user constraint-based techniques attempt to find knowledge that meet some sorts of constraints. There are two interesting concepts that have been used in user constraint-based techniques: metarules (Han and Kamber, 2000) and granule mining (Li et al., 2006). The aim of this chapter is to present the latest research results about data warehousing techniques that can be used for improving the performance of association mining. The chapter will introduce two important approaches based on user constraint-based techniques. The first approach requests users to inputs their metarules that describe their desires for certain data dimensions. It then creates data cubes based these metarules and then provides interesting association rules. The second approach firstly requests users to provide condition and decision attributes that used to describe the antecedent and consequence of rules, respectively. It then finds all possible data granules based condition attributes and decision attributes. It also creates a multi-tier structure to store the associations between granules, and association mappings to provide interesting rules.