Three-Layer Data Clustering Model for Multi-View Customer Segmentation using K-Means
Afgan Fazri Handoko1, Antoni Wibowo2
1Afgan Fazri Handoko*, Computer Science Department, BINUS Graduate Program–Master of Computer Science, Bina Nusantara University, Jakarta, Indonesia.
2Antoni Wibowo, Computer Science Department, BINUS Graduate Program–Master of Computer Science, Bina Nusantara University, Jakarta, Indonesia.
Manuscript received on February 12, 2020. | Revised Manuscript received on March 10, 2020. | Manuscript published on March 30, 2020. | PP: 1840-1846 | Volume-8 Issue-6, March 2020. | Retrieval Number: F7962038620/2020©BEIESP | DOI: 10.35940/ijrte.F7962.038620
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Customer Relationship Management (CRM) system is one of the methods to increase customer satisfaction with the services provided by the company. The data in a CRM system sometimes have not been utilized properly to find specific information about customer needs. The data mining process can help companies to segment and retrieve useful information about customers. The segmentation of customers can be categorized into groups based on the RFM (Recency, Frequency, and Monetary) values of the customers. Several studies have used the RFM model as a basis for customer segmentation. However, the methods proposed in previous studies are very specific to certain industries and the range of RFM scores used is also very subjective. Also, as the business grows there are challenges with RFM score measurement. RFM score measurement needs frequent adjustments in which this adjustment is not easy using the existing methods. Therefore, this study proposed a novel method to overcome the limitation of the existing methods using combined K-Means and Davies-Bouldin Index (DBI) to find the appropriate range of RFM scores. Based on our study in a telecommunication industry the proposed method simplify the measurement of the RMF scores as the data grows. This research also provided the appropriate RFM score range through the K-Means approach based on the optimal K value of the K-Means algorithm. Our proposed method could be implemented in other industries since it only depends on the values of RFM from the correspond data for each customer.
Keywords: Clustering, Data Mining, RFM Analysis, Segmentation.
Scope of the Article: Data Mining.