THE DESIGN AND IMPLEMENTATION OF TELECOM CUSTOMER SEGMENTATION USING DATA MINING TECHNIQUES
The aim of this research is to apply data mining techniques in telecom sectors to build models that can identify the contribution that customer makes to organization profitability based on the current relationship with the organization. The objective of this research is to design an enterprise customer segmentation model for Ethio Telecom that is used to identify the high value and behavior of enterprise customers. To meet the objective of the study we use a hybrid data mining process model, which consists of six phases to undertake the data mining process and to address the business problem. During the understanding of the problem, the business practices of the ET enterprise section are measured. This is done using interviews with business and technical experts and document analysis. Data preprocessing is done using different data mining methods. To prepare the data for analysis, we select 162315 records of customer data to conduct this research. After data preprocessing, we get 21126 records with thirteen attributes that are used for data mining tasks. This research is conducted using WEKA software version 3.8.2 and three clustering algorithms, namely, k-means, filtered and farthest first are used. Among clustering algorithms, the farthest first clustering algorithm has better clustering performance than other cluster algorithms (filtered and k-means). Hence, the model constructed by the farthest cluster is used to design a prototype. The result of this study is interesting and encouraging and confirmed that applying data mining techniques truly supports customer segmentation activities at ET. In the future, we recommended more segmentation studies by using a possibly large amount of customer records and employing other clustering algorithms to yield better results.
THE DESIGN AND IMPLEMENTATION OF TELECOM CUSTOMER SEGMENTATION USING DATA MINING TECHNIQUES, GET MORE COMPUTER SCIENCE PROJECT TOPICS AND MATERIALS