Design And Implementation Of E-commerce Customer Segmentation System Based On Cluster Analysis

Along with the network popularization, enterprises want to occupy a share in the field of electricity, customer resources is the core factor of victory. So enterprises start from product centered transfer mode to take the customer as the center, for the increasing and diversified trend of consumption data, customer segmentation of the traditional statistical method based on one dimensional attribute way has gradually exposed its shortcomings, and instead using appropriate Data Mining algorithm for amount and complex data information to achieve more accurate and effective customer segmentation, retention of existing customers and digging of potential customers, analyze and predict the future market trends, in order to achieve the differential marketing goals.In order to achieve the above objectives, this paper adopts clustering analysis method of Data Mining for some clothing electronic commerce enterprise to obtain the results of customer segmentation, specificly employing the improved K-means algorithm about neighbor transmission algorithm to realize the system design and implementation.By means of the introduction of the AP algorithm, the algorithm use clustering validity index in the process of iterative to carry on the supervision and guidance, to generate the neighbor propagation algorithm, implements the initialization of K-means algorithm. A new algorithm in this article is introduced in detail about the improvement process?implement process and verifying the the improved effective advantages of algorithm through experiments,embodied that the data set can effectively eliminate the noise of the point, and obtain more accurate initial clustering center, in order to improve the quality of clustering, at the same time, greatly improved the accuracy of clustering and tightness between each cluster. Compared with classic K-means algorithm,the improved K-means algorithm about the neighbor propagation algorithm can achieve more satisfactory clustering effect.The research work and achievements of this paper is as follows:1. This paper introduces the relevant theories of customer segmentation, includes it's concept, research background and significance, and detailed analysis of the customer segmentation used in the general method and the corresponding steps, through the detail of customer segmentation method, describing the perfection of the technology in the field of electricity and characteristics.2. This paper introduces the application of customer segmentation is the concept of data mining technology, the function and the commonly used method, and the clustering analysis algorithm of clustering of related principles and main methods for the overview of the system.3. The introduce of classic K-means algorithm and the AP algorithm,in detail in this paper, to analyze and provide the corresponding improvement about the K-means algorithm clustering criterion function and AP algorithm similarity measure.In order to achieve the K-means algorithm of the cluster number and cluster center initialization effectively, and improve the accuracy of clustering convergence condition, the improved K-means algorithm based on neighbor propagation algorithm is provided.Then the improved algorithm is applied to the experiment, compared with the former algorithm, achieving a better clustering result.4. For the design of e-commerce sites customer segmentation system, a detailed introduce of the e-commerce website customer demand analysis, containing the characteristics and the corresponding target system, describing the whole model of e-commerce sites customer segmentation system, to specific expound the design process about the data pretreatment and process of data mining.5. Completed the implementation of e-commerce sites customer segmentation system, the K-means algorithm is applied to a clothing e-commerce sites, to analyze and describe in detail the implementation process of the algorithm and performance analysis, and the segmentation results analysis, put forward the feasible marketing strategy.