Research On Supermarket Customer Relationship Management System Based On Agent

With economic development and social progressing, supermarket has been the main channel of everyday consumption of people. How to keep existing customers in the fierce competition has become the key to compete. However, nowadays, the research on intelligent decision of supermarket has been still at a primary stage, and it has been also few to analyze intensively and mining the data stored in the database. This paper based on the intelligent decision theory and studied in a deep-going way on supermarket customer churn and customer value, and established a system aimed at supermarket customer relationship management based on Agent.This paper studied intensively on how to establish the keeping strategies of closing drain customer through customer churn prediction model and customer value classification model. The major tasks in this paper describes as follows:(1)This paper imported Multi-Agent and data mining into supermarket customer relationship research, and realized communication mechanism of message pattern on JADE platform, and established a supermarket customer relationship management system based on Multi-Agent through mutual coordination of each Agent.(2)The system employed improved C4.5 decision making tree algorithm to predict whether customer would be loss. This algorithm imported weighted factor and weighted entropy of information on the basis of traditional C4.5 algorithm. In this way, it could not only adjust the distance between classified attribute and root node, but also make the characteristic of important sort more prominent. Meanwhile, rationality and validity on supermarket customer churn prediction has been proved through the analysis of theory and data.(3)The system used Ant Colony Algorithm based on Agent to classify the customers that prediction results was loss according to customer value, then divided customer value level through classified result and thus assisted supermarket manager to decide marketing strategy. Robustness of the module would become stronger through combining Agent and Ant Colony Algorithm, and the classified procedure and result were also more intuitive, autonomy and sociality of model was also enhanced.(4)The system was divided into customer churn prediction Agent, data intermediate handling Agent, and customer value classification Agent according to function, it realized communication among Agent through JADE platform and intellectualized harmony management among modules, degree of automation has also been raised in supermarket customer relationship management system.The research of direction of this paper was supermarket customer relationship management system, it expended the theory of system, implementation procedure and result. Supermarket customer keeping strategy was studied in depth from the standpoint of customer relationship management, thus it provided a basis for supermarket decider to decide marketing strategy of keeping customers. Meanwhile, the system had higher theory value and practical significance.