Mining Patterns of Select Items in Multiple Databases
A number of important decisions are based on a set of specific items in a database called the select items. Thus the analysis of select items in multiple databases becomes of primordial relevance. In this chapter, we focus on the following issues. First,
- PDF / 397,058 Bytes
- 20 Pages / 439.37 x 666.142 pts Page_size
- 48 Downloads / 208 Views
Mining Patterns of Select Items in Multiple Databases
A number of important decisions are based on a set of specific items in a database called the select items. Thus the analysis of select items in multiple databases becomes of primordial relevance. In this chapter, we focus on the following issues. First, a model of mining global patterns of select items from multiple databases is presented. Second, a measure of quantifying an overall association between two items in a database is discussed. Third, we present an algorithm that is based on the proposed overall association between two items in a database for the purpose of grouping the frequent items in multiple databases. Each group contains a select item called the nucleus item and the group grows while being centered around the nucleus item. Experimental results are concerned with some synthetic and real-world databases.
4.1 Introduction In Chapter 3, we have presented a generalized technique viz., MDMT: PFM+SPS, for mining multiple large databases. We have noted that one could develop a multidatabase mining application using MDMT: PFM+SPS which performs reasonably well. The following question arises as to whether MDMT: PFM+SPS is the most suitable technique for mining multiple large databases in all situations. In many applications, one may need to extract true non-local patterns of a set of specific items present in multiple large databases. In such applications, MDMT: PFM+SPS could not be suggested as it may return approximate non-local patterns. In this chapter, we present a technique that extracts genuine global patterns of a set of specific items from multiple large databases. Many decisions are based on a set of specific items called select items. Let us highlight several decision support applications where the decisions are based on the performance of select items. • Consider a set of items (products) that are profit making. We could consider them as the select items in this context. Naturally, the company would like to promote them. There are various ways one could promote an item. An indirect way of promoting a select item is to promote items that are positively associated with A. Adhikari et al., Developing Multi-database Mining Applications, Advanced Information and Knowledge Processing, DOI 10.1007/978-1-84996-044-1_4, C Springer-Verlag London Limited 2010
51
52
4
Mining Patterns of Select Items in Multiple Databases
it. The implication of positive association between a select item P and another item Q is that if Q is purchased by a customer then P is likely to be purchased by the same customer at the same time. In this way, item P becomes indirectly promoted. It is important to identify the items that are positively associated with a select item. • Each of the select items could be of high standard. Thus, they bring goodwill for the company. They help promoting other items. Therefore it is essential to know how the sales of select items affect the other items. Before proceeding with such analyses, one may need to identify the items that a
Data Loading...
