Risk Propagation Analysis Of Guarantee Network Over Big Data Technology

Recently,a large number of small and medium enterprises sprang out with the influence of the upsurge of the business startups.Joint guarantee became an important way for them to raise funds.The original intention is for better development,but the financial risk caused by the complexity of the guarantee network has become a problem that can not be ignored.In addition,the guarantee data expand rapidly with the development of the economy.Faced with massive data and complicated relations,the problem becomes more difficult.Based on the real data of a financial enterprise,this thesis uses complex network theory and big data technology to identify the factors that affect the risk propagation of the guarantee network quickly.Then,the corresponding measures for management and prevention are proposed to reduce the cost of risk control.The specific studies can be summarized as three aspects:(1)A community detection algorithm PFSCAN is proposed to deal with massive data based on MapReduce,which can identify the community structures,hubs and outliers quickly.Compared with SCAN,PFSCAN adds pruning strategy and parallel processing to improve the efficiency of the algorithm.Meanwhile,the experimental results show that the result of the community detection of SCAN and PFSCAN is the same,and the latter has good scalability when dealing with big data sets.(2)Based on the SEIR epidemic model,a comprehensive dynamic model for risk propagation in guarantee network is built,which is called dynamic model for risk propagation among enterprises with infective media and multistage infection.Infective media and multistage infection make the model more suitable for the environment of the propagation so that the analysis results are more accurate and provide a new perspective to deal with and control the guarantee network problem.(3)The numerical simulation of the dynamic model is carried out to analyze the influence of different parameters and risk sources on epidemic threshold and density of infected nodes.According to the conclusion,the corresponding measures are proposed to reduce the cost of risk management and solve the guarantee network problem.