Distributed Mining Model and Algorithm of Association Rules for Chain Retail Enterprise

Association rules discovery is the most important and demanding task of retail enterprises data mining. It has reflected the interesting related relation of products and is helpful to make correct decisions for retail enterprises. The paper describes distributed mining technology which is suitable for retail enterprise mining, the paper analyzes its distributed management system from three aspects: 1. the communication bandwidth, 2. branch storepsilas quantity 3. the characteristics of local database, and divides chain retail enterprises into two types in term of their distributed management systems. Define two kinds of retail enterprises; they are chain retail enterprise based on massed learning and chain retail enterprise based on dispersed learning. In order to satisfy mining task for two kinds of chain retail enterprise, the paper proposes two mining models and corresponding algorithms, finally do two contrast experiments under the simulation distributed setting. From the results of experiments, two algorithms ran well on corresponding chain retail enterprises.