Research On Distributed Data Mining Model Forchain Retail Enterprise

People's consumption concepts, patterns and purchase behavior have changed a lot with the the development of Information Technology and acceleration of economic globalization. Each node in chain retail enterprise has collected a lot of static and dynamic transaction data about customer,which form the sharing distributed data environment. So comprehensive decision-making based on distributed database to find hidden rules and knowledge is necessary.This Paper build the distributed data mining model named DDMMRCB to mine the distributed consumer behavior and consumption trends information in chain retail enterprise in order to improve the management and decision level and core competitiveness of chain retail enterprise.Firstly, considering the weakness of the distributed consumer behavior analysis theory, the distributed data mining model (DDMMRCB, Distributed Data Mining Model based on consumer behavior analysis) is proposed, which take the consumer behavior data in various distributed nodes as the data source, mobile agent operation platform as the framework.Three key technology in DDMMRCB model is heterogeneous data processing technology based on XML, improved distributed association rules DARMAIF algorithm (Distributed Association Rules Mining Algorithms based on Improved FP tree) and Distributed Neural Network Algorithm IDNNA (Improved Distributed Neural Network Algorithm).Secondly, the element mapping conversion rules and dynamic matching mechanism based on knowledge is proposed using the extensibility and ability of self-description of XML to mine heterogeneous data.Thirdly, the distributed association rules DARMAIF algorithm is proposed. The local original database site is decomposed and gradually form data subset composed of frequent 1 -items.Then sub-regional database is stored on the improved FP-tree structure. Through mining algorithms based on constrained sub-tree, the whole constraint-based association rules are obtained gradually.Fourthly, paper proposes an improved distributed Neural Networks IDNNA algorithm, which reduces the data dimensionality under the premise of effective information in the customer segmentation.Fifthly, the paper explores the application of DDMMRCB model in chain retail customer's consumer behavior analysis. Through the personal characteristics, customer consumption behavior and customer satisfaction factors to mine consumer behavior as correlation analysis, customer consumption patterns, customer classification which improve the retail chain enterprise's decision support and commodity management level have provided strong support.