The Research On Association Analysis And Its Application In Mobile Telecommunication Enterprise

Data mining is the procedure of extracting of implicit, original, useful knowledge in the database. Association analysis is one of the main technology in the research on data mining among a mass of theories and methods achieved. With the severe competition and market saturation in mobile telecommunication,the mobile telecommunication enterprises'marketing transfers gradually from the previous products-centered into the customer-centered. The main method of improving the profits of mobile telecommunication enterprise is to understand customers in-depth,guide customers, retain customers, enhance the value of existing customers, improve customers'satisfaction and lower customers'loss rate. Cross-selling is the important way of enhancing the value of existing customers, thereby it can increase the enterprise profits。For the needs of mobile telecommunication enterprise's cross-selling, this paper studies and proposes an algorithm for mining frequent pattern based on items-constraint and an optimized algorithm for generating association rules based on items-constraint, then establishes a cross-selling model for the mobile telecommunication enterprise and proposes marketing methods based on cross-selling。The research is summarized as follows:1.The technology of data warehouse and data mining is introduced briefly, and association analysis in data mining is disserted, involving the methods, characteristics and sorts of association analysis, with emphasis on the current commonly association analysis algorithms , such as FP-Growth mining algorithm based on FP-Tree and Apriori mining algorithm, and their advantages and disadvantages.2.Apriori algorithm and FP-growth algorithm can mine all frequent patterns which implied in the database, but the computation is too much when facing with the massive data in mobile telecommunication enterprise. Therefore, how to decrease computation is a major problem of frequent pattern mining algorithms. A method to solve the problem is only to mine the frequent patterns which are related to the particular business or item, rather than mining all the frequent patterns implied in the data in the mobile telecommunication enterprise cross-selling. It is regrettable that for such frequent pattern mining with specific constraints, the commonly used Apriori mining algorithms and the FP-Growth mining algorithm are not optimal. So there is the need to study and adopt a new mining algorithm。Based on above analysis, the paper proposes an ICFP-Tree(Items-Constraint Frequent Pattern-Tree) and a new ICFP-Mine (Items-Constraint Frequent Pattern-Mine) algorithm which directly mines in the tree。The ICFP-Tree compresses and stores all the information of the affairs included constraint items. ICFP-Mine algorithm directly mines the needed frequent patterns in the ICFP-Tree according depth-first strategy by adjusting the ICFP-Tree correlative nodes information, without any other additional data structure. Each time the algorithm mines just one sub-tree of ICFP-Tree and saves storage space effectively and greatly improves the efficiency of mining. Theoretical analysis and experimental results show that the ICFP-Mine algorithm is superior to Apriori and FP-Growth algorithm in memory occupancy and time costs. In addition, the paper introduces items-constraint into the conventional method of generating association rules from frequent items, proposes an optimized algorithm for generating association rules based on items-constraint.3.The paper discusses the basic theory, methods and process of cross-selling, studies an improved association analysis algorithm based on items-constraint applied in the area of mobile telecommunication enterprise's business cross-selling, then establishes mobile telecommunication enterprise's business cross-selling model and detailly discusses and analyses the results of the model's practical application.