Enhancing the Performance of Association Rule Generation over Dynamic Data using Incremental Tree Structures

Abstract

To discover a novel and dynamic approach for frequent itemsets generation and also for generating association rules is an imperative aspect in data mining. With the fast increase in databases, new transactions added, the incremental mining is acquainted to resolve the issues of maintaining association rules in updated databases. Earlier algorithms focused on this problem which consumed more time and costly to mine. This paper intends to analyze the tree construction like Frequent Pattern-tree(FP),PreOrderCoded(POC) tree and PrePostCoded(PPC) tree for sinking overheads and time constraints. To overcome theissueof updating association rules when new transactions addition this paper proposes a dynamic frequent itemsets mining approach using Incremental PreOrderCoded (IPOC)tree. This will reduce computational and scanning overheads of original dataset against addition of new transaction items and also works in an optimized way. An analysis was done on existing algorithms and compares time complexities for various standard datasets. The proposed approach shown better performance against existed ones over time and efficiency.