Segmentasi Pelanggan Menggunakan K-Means Clustering Berbasis Analisis RFM (Studi Kasus: Data E-Commerce PD XYZ)

10060221020, Jumidah
Suhaedi, Didi
Erwin Harahap

Bandung Conference Series: Mathematics
Vol. 6 No. 1 (2026): Vol. 6 No. 1 (2026): Bandung Conference Series: Mathematics

Abstract. Increasing business competition requires companies to gain a deeper understanding of customer behavior in order to maintain loyalty and minimize the risk of customer churn. This study aims to perform customer segmentation and identify customers who have the potential to churn using the K-Means Clustering method based on Recency, Frequency, and Monetary (RFM) values. The transaction data used in this study were obtained from PD XYZ, a store operating on an e-commerce platform in Indonesia. The data were processed through preprocessing and transformation stages, including a scaling process using RobustScaler to adjust the data scale and reduce the influence of outliers. The determination of the number of clusters was carried out using the Elbow method and evaluated using the Silhouette Coefficient and Davies–Bouldin Index. The evaluation results show that the number of clusters provides the best clustering quality, with a Silhouette Coefficient value of 0.771495 and a Davies–Bouldin Index value of 0.547348. The segmentation produced three customer groups: non-loyal customers with low activity and low transaction value, loyal customers with high activity and high transaction value, and moderate customers with medium activity and transaction value. Overall, 3,840 customers were identified as having churn potential, while 236 customers were classified as non-churn. The results of this study indicate that an RFM-based clustering approach is effective in understanding customer characteristics and can be used as a foundation for developing more targeted retention and marketing strategies. Abstrak. Peningkatan persaingan usaha menuntut pelaku usaha untuk memahami perilaku pelanggan secara lebih mendalam guna mempertahankan loyalitas dan meminimalkan risiko kehilangan pelanggan (customer churn). Penelitian ini bertujuan melakukan segmentasi pelanggan serta mengidentifikasi pelanggan yang berpotensi churn menggunakan metode K-Means Clustering berbasis nilai Recency, Frequency, dan Monetary (RFM). Data transaksi yang digunakan berasal dari PD XYZ, yaitu sebuah toko yang beroperasi pada platform e-commerce di Indonesia. Data diproses melalui tahap pra-pemrosesan dan transformasi, termasuk proses scaling menggunakan RobustScaler untuk menyesuaikan skala data dan mengurangi pengaruh outlier. Penentuan jumlah klaster dilakukan dengan metode Elbow, serta dievaluasi menggunakan Silhouette Coefficient dan Davies-Bouldin Index. Hasil evaluasi menunjukkan bahwa jumlah klaster memberikan kualitas klaster terbaik dengan nilai Silhouette Coefficient sebesar 0,771495 dan Davies-Bouldin Index sebesar 0,547348. Segmentasi menghasilkan tiga kelompok pelanggan, yaitu pelanggan tidak loyal dengan aktivitas rendah dan nilai transaksi rendah, pelanggan loyal dengan aktivitas tinggi dan nilai transaksi tinggi, serta pelanggan moderate dengan aktivitas dan nilai transaksi menengah. Secara keseluruhan diperoleh 3.840 pelanggan berpotensi churn dan 236 pelanggan tidak churn. Hasil penelitian ini menunjukkan bahwa pendekatan RFM berbasis klastering efektif dalam memahami karakteristik pelanggan serta dapat digunakan sebagai dasar penyusunan strategi retensi dan pemasaran yang lebih tepat sasaran.