PENGEMBANGAN SEGMENTASI PELANGGAN BERBASIS CUSTOMER LIFETIME VALUE DENGAN INTEGRASI TEKNIK DATA MINING TWO STEP K-MEANS CLUSTERING DAN ASSOCIATION RULE MINING (STUDI KASUS: KLINIK TUMBUH KEMBANG ANAK NIUMIU CDC)

• Main Author: Hardianti Santosa, Wini
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/76344
• Institution: Institut Teknologi Bandung

Customer Relationship Management (CRM) helps companies to get better understanding of customer needs and behavior. Customer needs and behavior insight is a key factor for gaining a competitive advantage by increasing customer satisfaction and fulfilling customer needs. One of the customer dimensions is customer recognition. Customer segmentation is a way for companies to understand customers better. In CRM applications, it is important to separate customers based on their value. Customer Lifetime Value (CLV) is a measure of the marketing division that projects customer value in relation to the company in a certain period. This study proposed a new approach by considering Customer Lifetime Value with the LRFMP (Length, Recency, Frequency, Monetary and Periodicity) model and integration of data mining techniques which are Two Step K-Means Clustering and Association Rule Mining (ARM) to perform segmentation and analysis of customer behavior on Niumiu CDC child development clinic. The LRFMP model was used as a basis for segmentation and determining customer’s CLV value. In this study, weight calculations were also carried out using the Analytic Hierarchy Process (AHP) for each LRFMP attribute which will then be used to calculate CLV for each segment. Then, LRFMP model was integrated with data mining techniques which are Two Step K-Means Clustering and Association Rule Mining. The data which was used in this study, come from customers transaction data in the Niumiu CDC child development clinic for 2 years. K-Means clustering step 1 has done to get initial segmentation from all data. After that, potential customer segments are determined for clustering step 2. K-Means clustering step 2 was carried out to get a better understanding of customers who provide good value to the company. Finally, Association Rule Mining is performed on each cluster based on K-Means clustering step 2 result to get deeper understanding of customer characteristics. Based on data processing result from 571 customers, 7 customer segments were formed, which are labelled as superstar, golden, dormant average value, dormant, new, new average value segment and new dormant segment. The measure of clustering goodness has been done based on internal validation using Davies-Bouldin Index, value for C2 is 1.05 and C0 is 0.79. In addition, association rules were also formed for each segment which provide deeper understanding of customer characteristics. This study approach provides better interpretation of each segment. Customer segmentation from this study can be used as a basis company consideration for decision making, determining business strategies, especially for marketing strategies and resource allocation.