Design And Implementation Of Business Risk Prediction Based On Process Mining

With the rapid development of the Internet,the number of web portals also showed explosive growth,the business of Electronic Commerce has become one of the most essential business in modern society.Meanwhile,with individual users to the depth of the Internet use,since the media personal portal websites also produced.With the business model update and increase,the electronic commerce business is becoming more complex,group purchase,distribution and other marketing methods emerge in an endless stream.Therefore,how to check the business model,and how to do business risk prediction has become the current problem.At present,more and more data is recorded in the event log,and these data contains business operation process.It is of great significance for enterprises to use these data to extract an important operation,find and fix the problem through the model.At the same time,the key to solve this problem is the business risk prediction and record the violation.Using the new technology process mining,through the platform of data information to build electronic commerce business process model in detail,and design a business risk prediction system to realize the intelligent control of the electronic commerce business management.The core requirement is mining user behavior process in this thesis.Firstly,mining and establishing a business process model from the user session event log.Secondly,by comparing the complexity and analyzing the realistic scene,find the most suitable algorithm to capture the business vulnerabilities.Similarly,compares the various Markov prediction model in accuracy and performance.And design the prediction algorithm based on the business problems.Using real data to verifying the accuracy of it.Then,analysis of system characteristics,role characteristic and research platform,using use case diagrams to describe the detailed requirements from both functional and non-functional aspects.The prediction system includes log file processing module,prediction model building module,prediction algorithm module,sending Email module,authorization management module and report module.Then,combined with the system structure,function module diagram,entity relationship diagram and database design,display the outline design of the project.And using function flowchart and class diagram,the detailed design of the project is described.Finally,author demonstrate the feasibility of algorithm by real data from and prove certain value of application and make an outlook on the future development of the project.After the launch of the project,accurately find the vulnerabilities in the electricity supplier business.Combined with Markov forecasting model and electricity supplier business,the prediction algorithm is applied to the prediction system.Since the on-line system,intercept illegal operations more than ten times,success reduce the number of illegal operations,and significantly reduces the business risk.