Sofyansah, Ridwan Saputra (2025) ANALISIS PERBANDINGAN ALGORITMA K-MEANS DAN HIERARCHICAL CLUSTERING UNTUK SEGMENTASI PELANGGAN DALAM OPTIMALISASI STRATEGI PEMASARAN MENGGUNAKAN METODE RECENCY, FREQUENCY, MONETARY (STUDI KASUS: PT XYZ). S1 thesis, Universitas Mercu Buana Jakarta - Menteng.
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text
41521010078-Ridwan Saputra Sofyansah-01 Cover - RIDWAN SAPUTRA SOFYANSAH.pdf Download (870kB) |
|
Text
41521010078-Ridwan Saputra Sofyansah-02 Bab 1 - RIDWAN SAPUTRA SOFYANSAH.pdf Restricted to Registered users only Download (456kB) |
|
Text
41521010078-Ridwan Saputra Sofyansah-03 Bab 2 - RIDWAN SAPUTRA SOFYANSAH.pdf Restricted to Registered users only Download (670kB) |
|
Text
41521010078-Ridwan Saputra Sofyansah-04 Bab 3 - RIDWAN SAPUTRA SOFYANSAH.pdf Restricted to Registered users only Download (676kB) |
|
Text
41521010078-Ridwan Saputra Sofyansah-05 Bab 4 - RIDWAN SAPUTRA SOFYANSAH.pdf Restricted to Registered users only Download (968kB) |
|
Text
41521010078-Ridwan Saputra Sofyansah-06 Bab 5 - RIDWAN SAPUTRA SOFYANSAH.pdf Restricted to Registered users only Download (465kB) |
|
Text
41521010078-Ridwan Saputra Sofyansah-08 Daftar Pustaka - RIDWAN SAPUTRA SOFYANSAH.pdf Restricted to Registered users only Download (335kB) |
|
Text
41521010078-Ridwan Saputra Sofyansah-09 Lampiran - RIDWAN SAPUTRA SOFYANSAH.pdf Restricted to Registered users only Download (1MB) |
Abstract
Di tengah meningkatnya persaingan bisnis khususnya di bidang retail, segmentasi pelanggan menjadi salah satu strategi kunci untuk mengenali kelompok pelanggan berdasarkan pola pembelian mereka. Tujuan dari studi ini yaitu untuk membandingkan efektivitas dan efisiensi algoritma K-Means dan Hierarchical Clustering dalam mengidentifikasi kelompok pelanggan, menggunakan pendekatan Recency, Frequency, dan Monetary yang didasarkan pada data transaksi pelanggan PT XYZ sepanjang tahun 2024. Studi ini menerapkan pendekatan kuantitatif menggunakan tahapan eksplorasi dan pembersihan data, perhitungan skor RFM, normalisasi data, penerapan algoritma clustering, serta evaluasi kinerja algoritma memakai metrik Silhouette Score, Davies-Bouldin Index, Cophenetic Correlation Coefficient, dan waktu komputasi. Temuan studi ini mengindikasikan bahwa metode K-Means mampu memberikan hasil nilai Silhouette Score sebesar 0,6505 dan Davies-Bouldin Index sebesar 0,4515 dengan waktu komputasi 0,0452 detik, sementara Hierarchical Clustering menghasilkan nilai Silhouette Score sebesar 0,6354, Davies-Bouldin Index sebesar 0,4647, dan Cophenetic Correlation Coefficient sebesar 0,8469 dengan waktu komputasi 46,6023 detik. Segmentasi pelanggan menghasilkan tiga kelompok utama, yaitu Best Customers (pelanggan dengan frekuensi tinggi dan kontribusi transaksi besar), Promising Customers (pelanggan dengan potensi tinggi namun aktivitas sedang), dan At-Risk Customers (pelanggan yang jarang bertransaksi dan memiliki kontribusi rendah). Dari hasil perbandingan, metode K-Means terbukti memberikan hasil yang lebih optimal dan efisien dibandingkan metode lainnya dalam menentukan segmentasi pelanggan berbasis RFM. Hasil segmentasi digunakan untuk menyusun pendekatan pemasaran yang lebih terfokus, seperti penyusunan program loyalitas bagi pelanggan bernilai tinggi serta strategi retensi untuk pelanggan yang berpotensi berhenti menggunakan jasa perusahaan, sehingga memperkuat proses pengambilan keputusan yang didasarkan pada analisis data di PT XYZ. Amidst the intensifying competition in the retail industry, customer segmentation has become a key strategy to identify customer groups based on their purchasing patterns. This study aims to compare the effectiveness and efficiency of the K-Means and Hierarchical Clustering algorithms in identifying customer segments using the Recency, Frequency, and Monetary approach, based on PT XYZ’s customer transaction data throughout 2024. A comparative quantitative approach is applied, including stages of data exploration and cleaning, RFM scoring, data normalization, clustering algorithm implementation, and performance evaluation using metrics such as Silhouette Score, Davies-Bouldin Index, Cophenetic Correlation Coefficient, and computational time. The findings of this study indicate that the K-Means method achieved a Silhouette Score of 0.6505 and a Davies-Bouldin Index of 0.4515 with a computation time of 0.0452 seconds, whereas the Hierarchical Clustering method yielded a Silhouette Score of 0.6354, a Davies-Bouldin Index of 0.4647, and a Cophenetic Correlation Coefficient of 0.8469 with a computation time of 46.8623 seconds. The segmentation process produced three main customer groups: Best Customers (highly active customers with significant transaction contributions), Promising Customers (potentially valuable customers with moderate activity), and At-Risk Customers (infrequent buyers with low contribution). Based on the evaluation, the K-Means method demonstrated more optimal results and faster performance, making it more suitable for RFM-based customer segmentation. The segmentation outcomes are utilized to design more targeted marketing approaches, such as loyalty programs for high-value customers and retention strategies for those at risk of churning, thereby enhancing data-driven decision-making processes at PT XYZ.
| Item Type: | Thesis (S1) |
|---|---|
| NIM/NIDN Creators: | 41521010078 |
| Uncontrolled Keywords: | Segmentasi Pelanggan, K-Means, Hierarchical Clustering, RFM, Retail Customer Segmentation, K-Means, Hierarchical Clustering, RFM, Retail |
| Subjects: | 000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 004 Data Processing, Computer Science/Pemrosesan Data, Ilmu Komputer, Teknik Informatika |
| Divisions: | Fakultas Ilmu Komputer > Informatika |
| Depositing User: | YOSUA EBENEZER PARDEDE |
| Date Deposited: | 04 Aug 2025 02:41 |
| Last Modified: | 04 Aug 2025 02:41 |
| URI: | http://repository.mercubuana.ac.id/id/eprint/96507 |
Actions (login required)
 |
View Item |
