Profile association rule mining from mixed database

The computerization of many businesses has led to massive amounts of data being stored in computers. The existing database system does not provide the user the necessary tools to extract useful knowledge from that data. Data Mining is the process of automatic extraction of novel, useful, and understandable patterns/models from large databases. Association rule mining is an important class of data mining that extracts interesting and frequent patterns from the data. The previous research in association rule mining used only transactional databases and previous association rule mining algorithms depended on the size of the itemset, the size of the dataset, and the size of the candidates set. Most of the developed association rule mining algorithms focused on extracting large itemsets efficiently by reducing I/O operations and the number of scans over the database, minimizing the set of candidate itemsets by pruning and paralleling and distributed the itemset generation process. The integration between transactional databases and relational databases is compulsory to extract profile association rule mining. This thesis presents a novel approach for discovering associations from a mixed database. The definition of a mixed database is introduced and integration methodology between transactional databases and relational databases in manual based is developed. This thesis deals with algorithmic and systemic aspects of profile association rule mining from integrated databases. The algorithmic aspect focuses on developing a new algorithm to extract profile association rules from a mixed database. We proposed two algorithms, CARMA and CARMAH, to extract a profile association rule from a mixed database in IF-Then format with illustrative examples. The systems aspect focuses on developing a scalable implementation for the designed algorithms. Extensive experiments have been conducted for testing the developed algorithms showing applicability and linear scalability with differently sized databases.