| The advanced management idea of"Take the customer as the center"manifests fully in the customer segmentation. Customer segmentation may meet customer need at whole hog, enable the enterprise to achieve the maximal profit. Data Mining is an advanced effective data analysis method. Its Application to customer segmentation will help the enterprises to gain their ends on the best way.As for customer segmentation in telecommunication enterprises, in this article I have mainly done the following work:Firstly, It introduced commonly used methods of customer segmentation in the domestic and foreign telecommunication enterprises, various characteristics of data mining algorithm by comparative analysis.With emphasis I analyzed the design concept of the classical K-means algorithm and the k-medoids algorithm, determined optimizing the initial value and the fixed segmentation standard and efficiency of carrying out to take the algorithm improvement as the key point, proposed tt-k-means and tt-k-mediods algorithms, and completed their detailed design.Secondly, I have gathered some sample data from Ganzhou Teleco mmunication Corporation, confirmed the system by some empirical values, and carried on the comparative analysis to the segmentation result.The experimental result indicated that segmentation results from tt-k-means and tt-k-mediods algorithms were efficient and effective, and the model produced was reasonable and understandable.Finally, I introduced the entire flow of data mining and the framework of data warehouse, and completed the customer segmentation system in the construction. The system took tt(two-times) algorithms as a core. It's composed of system administration module, pretreatment module, cluster analysis module, knowledge expression module. This system should become a subsystem to the business analysis system.The successful application of this system would facilitate to explore some other similar segmentation system. |
PDF Full Text Request