Research And Application Of Improved Association Rules And Clustering Algorithms

Today’s world,the wave of informatization using the Internet as a medium is booming,and data shows explosive growth.Data mining can extract valuable rules and models from various kinds of disorganized data,and its superiority makes data mining technology widely used in all walks of life.However,as data becomes more complex and diverse,it is of great significance to conduct more in-depth theoretical analysis and research on data mining algorithms at this stage and to conduct experiments on more data sets to verify their performance and enhance their value.The main research work of this paper is reflected in the following aspects:First,this paper introduces the definition,steps,algorithms,and data mining applications.Through case analysis or formula derivation,three classical association rule algorithms(Apriori algorithm,FP-growth algorithm,and Eclat algorithm)and three classical clustering algorithms(FCM algorithm,BIRCH algorithm,and two-step clustering algorithm)are analyzed.Then,the methods of pattern evaluation and cluster evaluation are introduced.Secondly,this paper proposes an improved association rule mining algorithm,the Gap-Apriori algorithm,to solve the problem that the Apriori algorithm cannot handle the hysteresis of the event association effect in time series data analysis.The algorithm combines the improvement ideas of time-domain compression,matrix reduction,and candidate set optimization strategy.It improves the algorithm’s efficiency by reducing the number of database visits and the generation of redundant candidate sets.Compared with similar algorithms,it is proved that it has higher operating efficiency.Aiming at the problems of a large amount of computation and low efficiency of the traditional hierarchical clustering algorithm,an improved hierarchical clustering algorithm PRI-MFC is proposed.The algorithm improves the hierarchical clustering algorithm from multi-stage clustering and divides the clustering analysis process into two stages: fuzzy pre-clustering and Jaccard fusion clustering.The algorithm can identify the cluster density,reduce the algorithm’s computational complexity,and improve the operating efficiency of the algorithm.The improved algorithm is compared with the clustering results of similar algorithms on artificial datasets and UCI datasets,and it is proved that it has a better clustering effect.Finally,this paper explores the application of the improved algorithm.The improved association rule algorithm is applied to the historical transaction data of China A-shares,and the linkage law of China A-shares in recent years is obtained.The improved hierarchical clustering algorithm is applied to Chinese A-shares’ value and price cluster analysis.Some high-value stock groups and the price distribution of individual stocks are obtained.In addition,a construction and management strategy for the investment stock pool is proposed through the combined application of two improved algorithms.The rules mined by the algorithm can provide decision support for investors’ in-stock selection and have specific practical application value.