Research On Application Of K-means Method In The Customer Management Project Of D Logistics Company

Customer management is the enterprise in a clear strategy,business model and specific market,according to the customer’s attribute,behavior,needs,preferences and values and other factors for the classification of customers,and provide targeted products,services and marketing mode.Enterprises that can correctly assess the value of customers can provide personalized service to different customers,and increase the profit of the enterprise while effectively managing the customer relationship management.Knowledge discovery and data mining technology has a wide range of applications in customer relationship management,the appropriate knowledge discovery and data mining tools have an important supporting role for customer relationship management.First of all,the D logistics company telemarketing customer management project reference CRISP-DM(cross industry standard process for data mining)model,the project is defined as 6 processes: business understanding,data understanding,data preparation,modeling,evaluation and deployment.Secondly,according to the objectives of the project,the paper adopts the method of literature survey and expert survey to find the relevant indicators of customer management.After the data collection,cleaning,conversion,we completed the missing value completion,and delete the coaxial relation factors.Next,by normalizing all the factors,we get the standard factor data.Then,in view of the above factors,the kmeams method is used to model classification,and the value of K in modeling is established,and it is concluded that the classification is divided into 4 kinds,which can minimize the radius of clusters.We finally get a list of customer details in each category after the k-means classification.Finally,through the analysis of the business volume and the sales call,the validity of the model is analyzed and verified.And according to the characteristics of all kinds of customers,we put forward the management method for all kinds of customers.In addition,for the 3 months of new customers,the use of KNN methods for new customers to predict classification,to ensure that new customers can be added in a timely manner to the unified classification of the customer complete system management.