基于K-Means聚类算法的客户细分分析-应用数学专业论文.docxVIP

Research on Customer Segmentation based on K-Means clustering algorithm ABSTRACT Wit h the advance ment o f techno logy and the develop ment o f the nat iona l economy, the communicat ions ind ustry has got rapid developme nt. The telecommunicat ions companies ha ve gradua lly rea lized that customers are the really va luab le assets, and have changed their marketing model fro m mass–oriented market ing into one- oriented marketing. The most important in establishing customer- centric marketing s ystem is to segme nt custo me rs, i.e., classify custo mers into categories based on customer characterist ics mined fro m raw custo mer data set so that enterprises can provide custo mers wit h exc lus ive products and services. In this thes is, based on traditiona l custo mer segmentat io n theor ies, K- means clustering algor ithm has been studied and has been applied into communicat ion enterprise’s customer segmentation. The main works of this thesis are as follows: It introduces the basic concepts, theories and methods of c ustomer segme ntat ion, a nalyzes its curre nt research at ho me and abroad, and focuses on the research on the features and major technica l methods of custo mer segmentat ion in the enterpr ises. It also introduces the basic concepts, princ ip les and releva nt evaluat ion cr iter ia of c luster ana lys is, studies so me popular ca lculat ion met hods, and focuses on the working principle and characteristics of K - means algorithm. Based on K- means a lgor it hm,a nove l c us former se gment met hod has been designed.First ly, it collects customer data from domest ic communicat ion enterpr ise , and disposes these data according to the character ist ics of consumer data in t his enterprise, then discusses and designs processing met hods inc luding data cleaning, noise remo ving, and outlier. Secondly, it ana lyzes the dime ns ion characterist ic of customer data to reduce the data dimens ion, whic h are by the way of analys is ing the