Applied Research Of Hybrid Genetic Clustering Algorithm In Customer Segmentation

The fierce market competition is attributed to the competition of customer resources,80% of enterprise profit is earned from 20% of their customers, how to manage and make use of the client resources is the main content of customer relationship management. Customer segmentation, as the key factor of implementing the customer relationship management, its purpose is to divide the clutter client resources into clear clusters, which can be used for targeted business service to the very customer by enterprise.In this paper,we use cluster analysis to mine the client resources, and the classical k-means clustering algorithm has good performance on it, but it's sensitive to the initial selection of cluster centers:different cluster centers will lead to different clustering results and accuracy, for this problem, this paper presents a hybrid genetic clustering algorithm, and test the algorithm's performance by a series of experiments.The main research work of the paper includes:in the context of maintaining the excellent framework of genetic algorithm and k-means algorithm, we encode the data in the way of floating-point, which is clear and intuitive, rather than the traditional binary coding.The change of coding way leads the selection operator, crossover operator, mutation operator to improve accordingly.In addition, due to the importance of the initial cluster centers, this article adopts a uniform approach of the clusters rather than the random selected clusters. Finally, we mine two group datasets in both the hybrid genetic clustering and traditional k-means cluster algorithm, and we compare the results of the two methods.