Design And Implementation Of Trade Mining And Prediction System Based On Counter Logs

In order to improve the operating efficiency and service quality of banks,technologies such as data mining and machine learning have been widely used in data analysis of the banking system.Bank counter is an important transaction channel for the daily operation and management of commercial banks.The daily operation of bank counter system generates a large number of operation logs,which contain the business information and potential business needs of customers.By analyzing counter logs,on the one hand,the sequence relationship between trade operations can be mined,which helps the software development and testing engineers of the counter system obtain the actual business scenario;on the other hand,it can help construct the trade prediction model,which is used to predict the next trade operation and assist the teller to meet the customer's business needs quickly and accurately.To this end,the thesis has designed and implemented a trade mining and prediction system based on counter logs,focusing on the study of order relationships between trade operations in actual business scenarios,such as causal and parallel relationships,in order to achieve more effective results of trade mining and prediction.The main tasks of this thesis are as follows:(1)After analyzing the structure of the counter log of the rural commercial bank,the system uses a script to extract the trade trajectory.The system uses Prefix Span algorithm to mine the frequent sequence pattern of the trade trajectory sequence,which screens out the trade operation sequence with potential connection.(2)The mining of relationships between trades is completed by constructing Petri workflow networks.The results are stored in the Neo4 j graph database.The consistency checking tool Cobefra is used to evaluate the mining results,which verifies the effectiveness of using process mining method to solve the trade mining problem.(3)A Compact Prediction Tree(CPT)algorithm that integrates relationships between trades is designed,and the experimental comparison with the original CPT algorithm,Markov algorithm and Markov algorithm integrating the inter-transaction relationship proves that the proposed algorithm has a better prediction effect.(4)On the basis of the researchs above,the trade mining and prediction system based on counter logs is realized.Spring Boot framework is used to realize the basic functions of the system.Apache Shiro framework is used to guarantee the security of the system.Quartz framework is used to complete the timed task scheduling.The core function modules include data management module,system management module,trade mining module,trade prediction module and timed task management module.(5)The function and performance of the system are tested.The system successfully provides the trade relationship network diagram for the bank technical personnel through mining counter logs.The system can provide accurate trade prediction results for the counter business system for reference.