Comparative Analysis of Association Rule Mining on Basis of Frequent Pattern Tree

Abstract

There are methods for obtaining valuable data from the vast databases used in engineering, marketing, finance, and other fields. Finding this important data is done through data mining. Association Rule Mining: important data is extracted using a widely utilised data mining technique using the FP tree data structure. Rules or correlations are criteria that can be used to describe useful information. Considering large number of data elements, association rule mining uncovers intriguing linkages and/or correlation links such that the detailed analysis can take place on basis of statistical regularity. Association rules highlight attribute value situations in a dataset that commonly co-occurs. These types of data are sent by association rules as "if-then" expressions. In contrast to if-then logic rules, association rules are based on facts and are probabilistic. Apriori algorithm is used to generate candidates from transactional databases and time series databases for testing. Apriori employs a "bottom-up" methodology in which frequently used item sets are expanded. However, creating candidate sets is expensive, particularly when there are many or lengthy patterns. We can discover rules and relationships between pattern elements with the use of patterns. Transactional databases need to be optimized since they contain a lot of rows and columns from various sources, including banks, shops, engineering equipment, and more. The frequent-pattern tree (FP-tree) is an extended prefix-tree structure for storing compressed information about frequent patterns. From this enormous dataset, the FP tree will look for potential patterns. Rules are developed for a dataset and implemented where improvement is needed with the aid of the support and confidence qualities of FP trees.