Research Of Clustering Based On Swarm Intelligence And Apply To Analyze Stock Plate

In the background of big data,the data of stock market are generating more and more.Traditional analysis methods can't meet the needs of modern stock plate analysis.How to dig out useful information from the massive stock market,and analyze the stock market has become an important issue.In this environment,data mining develops rapidly as a new data technology which has made great contributions to the development of information resources.Clustering is an unsupervised learning data mining technology,which can analyze the internal characteristics and the relationship of the data.To avoid facing the huge data set,Clustering explore the potential link between data mining and features through the formation of a number of clusters.Swarm Intelligence algorithm is a kind of evolutionary computation technology which has the advantages of simple implementation,fast convergence speed and global convergence.It has been have caught more and more attention of experts and scholars at home and abroad.Swarm intelligence optimization algorithm is an important branch of artificial intelligence and contact with artificial life and genetic algorithm closely.Swarm intelligence optimization algorithm simulates the behavior of natural biological groups and uses the information transfer among individuals and cooperation to achieve the purpose of optimization.Fruit fly optimization algorithm(FOA)is kind of Swarm Intelligence algorithm and has been widely used in many fields because of its high precision and few parameters.But FOA still has its drawbacks that FOA was not effective when dealt with complex optimization problems.In order to improve the situation this paper proposed an improved algorithm named real-time learning fruit fly optimization algorithm(RTLFOA).The RTLFOA acquired population knowledge from searching in real time.The learned knowledge guided algorithm mutation that fruit flies updated their location with different purposes.The fruit flies with bad smell concentration would search the area more broadly to improve the convergence speed and the others would search the surrounding area to improve the convergence precision.According to the application in several benchmark functions and comparison among different algorithms,RTLFOA showed an effective improvement in its convergence speed and convergence accuracy.The affinity propagation algorithm is an unsupervised clustering algorithm,which does not need to specify the number of clusters and has good clustering results when dealing with large data sets.However,the deviation parameters of AP clustering algorithmcan't be set reasonably.(1)Based on the global search ability of Swarm Intelligence optimization algorithm,a Fruit Fly Optimization algorithm based on differential distance(FOADD)is proposed.To avoid the invalid movement and improves the search efficiency of the algorithm,the FOADD algorithm makes the fruit flies have the same distance from the origin move at the same time.(2)Based on the fast searching ability of FOADD,an affinity propagation algorithm based on the FOADD is proposed.The deviation parameters are optimized in the search space.Through the testing by classical data sets,the validity is proved According to the Silhouette validity index and clustering results.Using the AP-FOADD clustering algorithm analyzes the stock market.In this paper,112 listed companies in Beijing were selected.Four financial indicators that reflect the profitability of listed companies as the research object to classify the companies.According to the clustering result to analysis the management ability and the development potential of listed companies.According to the annual reports of listed companies,the improved algorithm is proved to be effective.The results provide a theoretical basis for investors to invest and avoid investment risks.