Yulia Indriani
Universitas Teknokrat Indonesia

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

Perbandingan Algoritma K-Means dan DBSCAN dalam Pengelompokkan Data Penjualan Elektronik Tria Setyani; Yulia Indriani; Muhammad Fadli; Erliyan Redy Susanto
Progresif: Jurnal Ilmiah Komputer Vol 21, No 2 (2025): Agustus
Publisher : STMIK Banjarbaru

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35889/progresif.v21i2.2775

Abstract

This study aims to compare the performance of the K-Means and DBSCAN algorithms in clustering electronic sales data at CV Rey Gasendra. Clustering methods are used to identify sales patterns based on the transaction intensity of each product. The K-Means algorithm was implemented by determining the optimal number of clusters using the Elbow Method and Silhouette Score evaluation. Meanwhile, DBSCAN was applied using a density-based approach, which does not require specifying the number of clusters in advance and relies on eps and minPts parameters. The results showed that K-Means produced more stable and interpretable clusters with a Silhouette  Score of 0.961, whereas DBSCAN excelled in detecting outliers and handling irregular data distributions, despite generating a large amount of noise. Performance evaluation using metrics such as Silhouette  Score, Davies-Bouldin Index (DBI), and Sum of Squared Errors (SSE) indicated that K-Means is more suitable for structured product segmentation, while DBSCAN is more effective for exploring unusual sales patterns. These findings can support data-driven business decisions, such as marketing strategies, inventory management, and product recommendations.Keywords: K-Means; DBSCAN; Clustering; Sales data; Product segmentation AbstrakPenelitian ini bertujuan untuk membandingkan kinerja algoritma K-Means dan DBSCAN dalam pengelompokan data penjualan elektronik pada CV Rey Gasendra. Metode clustering digunakan untuk mengidentifikasi pola penjualan berdasarkan intensitas transaksi setiap produk. Algoritma K-Means diimplementasikan dengan penentuan jumlah klaster optimal menggunakan metode Elbow dan evaluasi Silhouette Score. Sementara itu, DBSCAN diimplementasikan dengan pendekatan berbasis kepadatan, tanpa memerlukan jumlah klaster awal, dan menggunakan parameter eps serta minPts.  Hasil penelitian menunjukkan bahwa K-Means memberikan hasil clustering yang lebih stabil dan mudah diinterpretasikan dengan Silhouette Score mencapai 0,961, sedangkan DBSCAN unggul dalam mendeteksi outlier dan menangani data dengan distribusi tidak beraturan, meskipun menghasilkan banyak noise. Evaluasi performa berdasarkan metrik seperti Silhouette Score, Davies-Bouldin Index (DBI), dan Sum of Squared Errors (SSE) menunjukkan bahwa K-Means lebih cocok digunakan untuk segmentasi produk penjualan yang terstruktur, sedangkan DBSCAN lebih efektif untuk eksplorasi pola penjualan yang tidak lazim. Temuan ini dapat membantu pengambilan keputusan bisnis berbasis data, seperti strategi pemasaran, manajemen stok, dan rekomendasi produk.Kata kunci: K-Means; DBSCAN; Clustering; Data penjualan; Segmentasi produk
Perbandingan Kinerja Algoritma K-Means Dan Dbscan Dalam Segmentasi Nasabah Berdasarkan Data Pemasaran Bank Yulia Indriani; Tria setyani; Heni Sulistiani
Management of Information System Journal Vol 3 No 3: Juli 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/mis.v3i3.2119

Abstract

Customer segmentation is one of the important approaches in the banking industry to improve the effectiveness of marketing strategies and risk management. This study aims to compare the performance of two commonly used clustering algorithms, namely K-Means and DBSCAN, in segmenting customers based on bank marketing data. The data used comes from the bank marketing dataset available on Kaggle, including attributes such as age, type of employment, marital status, education level, savings balance, and previous campaign history. The analysis process includes data pre-processing, feature extraction, data standardization, and implementation of the clustering algorithm. Evaluation of the results using the Silhouette Score metric shows that the K-Means algorithm produces a higher Silhouette Score value and a lower Davies-Bouldin Index (DBI) than DBSCAN indicating that K-Means is able to produce more solid (compact) clusters and has clearer boundaries between data groups. However, DBSCAN shows advantages in the ability to detect noise and group data with irregular shapes. Thus, the selection of the best algorithm is highly dependent on the characteristics of the data and the purpose of the segmentation. For bank marketing data with a relatively uniform distribution, K-Means is a more optimal choice.