Research On Intelligent Analysis Of Customer Characteristics In Retail

With the develepment of large retail stores and the arrival of foreign enterprises, retail stores in China meet serious competition in the retail .If retail enterprises want to be in an invincible position in the competition, must give full attention to customer relationship management.The exponential growth in data and development of intelligence analysis technology has brought opportunities and challenges to customer relationship management.Customer segment and finding out the most valuable customers, the integration of limited resources to customers' differentiated services, paly an important role in remaining competitive powerin the brutal competition . And it is the key to have all the characteristics of clients when finding out the most valuable customers.Data Mining can efficiently deal with the large number of historical and current data, from the database can find some potential, useful and valuable information for the retail stores.This paper aims to use the principles of data warehouse and data mining methods toretail enterprise customer segments, and then to make customer characteristic analysis. Start with the special nature of the retail industry, with customer relationship management theory and data mining techniques and theory,study a large retail supermarket , build customers management system based on data warehousing and data mining .The major work is as follows:(1) In this paper, the object of study is supermarket transaction and customer data, there are more redundant data, invalid and wrong data and fewer cuostmer features in Chinese supermarkets, to deal this problem, this paper design a new data cleansing and attributes Extraction methods and use Transact-SQL to achieve, make a solid foundation for customer segmentation and customer profiling.(2) Data-processing software have their own merits,this paper based on the practical application , according to data and software features, integrate the sofeware from the pretreatment, data warehouse building, Construction of model, model evaluation processes. Give full play to the advantages of a variety of software.(3) In this paper, we give a new customer segments method, by empirical, through cluster analysis approach to customer segments. We use Distance Between Cluster method to achieve a theoretical foundation and data unification.In order to compare the traditional methods and clustering methods, we use Sum of Squares Between Groups method and Sum of Squares in Groups method.Confirmed the advantage of clustering method.(4) Based on customer segments,in order to mine different groups of customer characteristics,use Association Rules to different groups of customer and get rules about customer characteristics,and then give interpretation of the rules according to the actual work of supermarket,finally give reference to the supermarket's marketing and management work.(5) According to customer preferences on the level of discount merchandise, using mathematical statistics method, make division of the clients in another direction, and then use Association Rules, finallly get characteristics of the various types of customers which have different preferences of discount.The paper provides some methods and results which have certain, aimed at helping retail supermarket to make accurate customer segments, thereby comprehensively grasp the features of various types of customers, improve marketing development and marketing planning, and finally improve the competitiveness.