KLUSTERISASI DATA MINING SEGMENTASI PELANGGAN NETFLIX MENGGUNAKAN METODE K-MEANS DENGAN RAPIDMINER

Abstract

The intense competition in the streaming industry has pushed Netflix to understand its customer base more deeply. This study aims to segment Netflix customers using the K-Means Clustering method to identify customer characteristics and behavior patterns. The analysis was carried out using the RapidMiner Studio tool, utilizing the Netflix Userbase dataset for the period June-July 2023, which consisted of 2500 customer data. The variables used include Subscription Type, Monthly Revenue, Country, Age, Gender, and Device. The results of implementing K-Means Clustering with k = 5 produced optimal customer segmentation, as evidenced by the Davies Bouldin Index value of -2.920. The analysis identified five main clusters: new customers with high engagement levels (Cluster 0), loyal old customers (Cluster 1), active customers with regular payment patterns (Cluster 2), dominant customers from the United States (Cluster 3), and a special segment of German customers (Cluster 4). The model evaluation shows good cluster cohesion with optimal average within centroid distance values ​​for each cluster. This study recommends Netflix to develop a more personalized marketing strategy based on the characteristics of each cluster, and to conduct segmentation analysis periodically to follow changes in customer preferences. The results of this segmentation can be the basis for strategic decision-making in developing more targeted services and marketing.