Association rule mining system based on improved Apriori algorithm

Abstract

The invention provides an association rule mining system based on an improved Apriori algorithm. The association rule mining system comprises a data pretreatment module, a connection module, a pruning module, a frequent item statistics module and an association rule generation module, wherein the data pretreatment module interacts with a database, and takes charge of converting text data in the database into an integer format capable of carrying out bit operation; the connection module, the pruning module and the frequent item statistics module are used for forming concrete realization of the Apriori algorithm, and taking charge of regeneration of a frequent item set; and the association rule generation module interacts with the frequent item statistics module, and takes charge of converting frequent items generated by the frequent item statistics module into specific association rules. By virtue of a frequent item statistics method based on bit operation, the complexity of pruning operation is simplified; and the database scanning frequency is reduced, so that the association rule mining efficiency is improved; the consumption of system resources is reduced; the relatively efficient and convenient association rule mining business can be provided for enterprises and merchants; and the association rule mining system has the relatively great practical value.

Description

technical field [0001] This invention relates to a set of association rule mining system based on improved Apriori algorithm, which is an association rule mining system based on bit operation for frequent item statistics. In the process of realizing Apriori algorithm, the system reduces the number of candidate item sets in the pruning operation. The number of inspections of subsets and reducing the number of database scans when frequent itemsets are generated from candidate itemsets belong to the field of data mining. Background technique [0002] With the successful application of the "beer, diaper" case in the marketing field, people are more and more interested in mining association rules in the continuously collected data, and the application range of association rules has gradually expanded from supermarket marketing to more fields. The current association rule mining algorithms mainly include Apriori and FP tree. Due to the limitations of large memory consumption, com...

AI Technical Summary

Problems solved by technology

Due to the limitations of large memory consumption, complex implementation, and high system requirements, FP trees have not been widely used in actual use. Currently, they are mostly used as teaching cases or research topics.

The Apriori algorithm is actually the most used method in the association rule mining process, but this algorithm still has problems such as many database scans, large candidate sets, and low mining efficiency.

Images