<![CDATA[Pandhawa Sejahtera Dropship is a business in the dropship service sector. This business faces challenges in developing targeted marketing strategies due to the absence of transaction data-based customer segmentation. This study aims to implement customer segmentation using the Recency, Frequency, Monetary (RFM) method and compare the K-Means and Fuzzy C-Means algorithms in grouping customers based on transaction data. The transaction data used amounted to 2,376 records over a one-year period. The methods applied include data preprocessing, RFM value calculation, Min-Max normalization, determination of the optimal number of clusters using the Elbow Method and Silhouette Score, and cluster quality evaluation using the Davies-Bouldin Index (DBI). The results showed that the optimal number of clusters is 3 (k=3) with a Silhouette Score of 0.6242. The three clusters formed are: Cluster 0 (Champions) with 362 customers (25%) characterized by low recency (45.2 days), high frequency (3.8 times), and high monetary (Rp 1,256,780); Cluster 1 (Regular) with 724 customers (50%) characterized by moderate recency (215.3 days), low frequency (1.2 times), and moderate monetary (Rp 345,670); and Cluster 2 (At Risk) with 362 customers (25%) characterized by high recency (345.6 days), very low frequency (1.0 times), and low monetary (Rp 124,890). Based on the method comparison, the K-Means algorithm produced a DBI value of 0.77 and a Silhouette Score of 0.54, better than Fuzzy C-Means with a DBI value of 1.05 and a Silhouette Score of 0.39. Thus, the K-Means algorithm is declared as the best method for customer segmentation on Pandhawa Sejahtera Dropship transaction data. These segmentation results can serve as a basis for developing more targeted and efficient marketing strategies.]]>
Copyrights © 2026
