The Optimization Of Frequent Feature Mining And The Improvement Of Correlation Classification Method

Aiming at the problems of traditional association classification algorithms,such as the difficulty of rule updating under the condition of huge data volume increase and the need for optimization and extraction of rule redundancy,this paper proposes an improved association classification algorithm based on bayesian distribution dynamic update rules and genetic algorithm optimization rules.The main research points are as follows: 1.Based on the distribution characteristics of the item variables,associated classification frequent set mining based on frequency calculation is turned into a study based on the distribution of probability calculation and comparison problems.Research on the updating of association rules is turned into the research of mining frequent sets and updating rules with bayesian distribution of item variables under new sample data.The prior distribution of distribution parameters of item variables and the sample distribution of new data are used to obtain the posterior distribution of distribution parameters of item variables dynamically.The post-test distribution supports frequent set mining and rule updates,making frequent sets and rules mining has the advantages of simple,dynamic,and informative diversity.2.For the extracted association rules with random uncertainties,the optimization of the rules is studied: K-means clustering is performed by transforming the rules into a vector sample representation of the random possibilities of multiple premises and conclusions.The number of clusters and clustering results,the selection of the optimal clustering center,and converted to a representative optimal rule,get the result of the algorithm.Matlab programming to achieve the optimization process,and verify the operating results with the test subset.3.The method of distribution-based frequent feature mining,dynamic update rules based on bayesian distribution,and genetically optimizing association rules proposed in this paper are applied to the university asset efficiency management project.Satisfactory results are obtained,and the feasibility of the improvement and application of this method is demonstrated effectiveness.