Research On Association Rules Mining Algorithm Based On Particle Swarm Optimization

Association rule mining is one of the most important branches of data mining,and its main purpose is to find out the hidden relationships or links in the database.With the arrival of the era of big data,the problem of traditional association rule mining algorithm is increasingly obvious.As a consequence,the efficiency of the algorithm declines.As a representative of swarm intelligence optimization algorithm,particle swarm optimization has been widely used in different fields,such as the analysis of association rules,in recent years.In this thesis,by combining the particle swarm optimization algorithm and the association rule mining algorithm,the efficiency of association rule mining algorithm has been improved.In order to let association rules change with time,the gray model of particle swarm optimization is used for predicting the support and confidence vector of dynamic association rules,helping the decision makers grasp the development of the situation and providing a reference for making decisions.In order to make better association rule mining algorithm research,on the basis of a lot of literature and analysis on the present situation domestic and overseas,some problems existing in this area are found out,which puts forward the main content of this thesis.First of all,the basic concepts and principles,classification,classical algorithm and improved algorithm of association rules are introduced,which helps in understanding the purpose and significance of association rules mining.Then,the definition of dynamic association rules and the idea of algorithm will be given out as to understand the difference between dynamic association rules and association rules.Finally,the principle,the steps and the comparison between genetic algorithm and particle swarm optimization algorithm will be analyzed,providing the basis for the combination of particle swarm optimization algorithm and association rules algorithm.To address the declining efficiency in dealing with large database mining by Apriori algorithm,a method of second order particle swarm optimization algorithm based on the association rule mining is proposed.Firstly,based on the principle that each partition can be put into memory,the original database is divided into n non-overlapping sub-database using Partition algorithm.Secondly,we do association rule mining in each sub-database with Apriori algorithm,and then obtain the best rules set by using second order particle swarm optimization algorithm,extracting some valuable rules that are easily overlooked.Lastly,we will merge these rule-sets generated from each sub-database globally and calculate the actual support and confidence.The algorithm can not only reduce the number of scanning database,but also can extract the association rules which are ignored by a single reference standard.Through the implementation of the algorithm on the Matlab platform,the comparativeexperiments are carried out on different data sets,and a lot of similar algorithms are compared.The experiment shows that the algorithm is feasible and effective.According to the analysis and prediction of the trend of support vector and confidence vector in the mining of dynamic association rules,an improved gray model of particle swarm optimization with buffer operator is proposed.The Algorithm,by using the improved particle swarm optimization algorithm,joins two search mechanism to improve the local search ability of the algorithm,and optimizes background values of the gray model at different times,improving the prediction accuracy of gray model.The algorithm can not only find the rules with the change of time,but also predict the development trend of the rules,providing reference for decision-making.Through the implementation of the algorithm on the Matlab platform,the prediction accuracy of different algorithms is compared.The experimental results show that the prediction accuracy of the proposed algorithm is up to the first level,and it can meet the normal requirements.A series of contrast experiments have been carried out to prove the feasibility and effectiveness of the proposed algorithm,but it still needs some experiments in practical application.In this thesis,we select the data of the migration census to analyze the association rules,first select the characteristics of inter provincial flow as the basis,to analyze the characteristics of inter provincial migrants,such as age,nationality,household registration type and education level,then makes the mining of association rules for the reasons of the flow of inter provincial floating personnel,and gets the characteristics of the flow.With the analysis,we are going to provide constructive advice for the relevant departments in terms of strengthening the management of personnel,and,from the mining results,demonstrate the actual value and significance of the improved algorithm,to ensure the rigor of the algorithm.