Multi-objective hybrid intelligent optimization based model development for automatic rule mining in numerical data

Discovery of association rules in large data sets is one of the most important tasks of data mining. Almost all of the methods used for mining association rules are proposed for discrete-valued data sets. For this purpose, in many real-world data that have numeric-valued attributes should be transformed into binary or discrete-valued in order to be used by the classical rule mining algorithms. However, this a priori discretization process changes the real data and real high-quality rules cannot be discovered from the changed or modified data due to data loss and attribute interactions. Automatically adjusting the attribute intervals at the time of the mining process using the same unique rule mining algorithm without a preprocess such as discretization is more meaningful. In this thesis, differential evolution and sine-cosine algorithm based novel hybrid multi-objective evolutionary optimization methods are proposed for rapidly and directly mining the reduced high-quality numerical association rules by simultaneously adjusting the relevant intervals of related attributes without finding the frequent itemsets. These algorithms perform a global search and find the high-quality rules set in only one execution by modeling the rule mining task as a multi-objective problem that simultaneously meets different conflicting metrics. The algorithms proposed in this thesis study ensures the discovered rules to have high confidence and support and to be comprehensible. They also automatically find the related minimum intervals for the attributes of the mined rules. Further, the proposed methods automate the rule mining problem by eliminating the need for metrics such as minimum confidence and minimum support determined beforehand for each data set. The performance of new algorithms proposed in this thesis were tested with the state-of-the-art methods on real data sets. The results show the superiority of the proposed methods on the data sets containing fewer attributes and higher number of instances.