<![CDATA[The rapid development of information technology has changed the way products are sold, especially through online platforms that are increasingly in demand. In increasingly tight business competition, companies need to understand the differences in customer needs and behavior. Inability in this regard can make it difficult to design effective marketing strategies. Therefore, customer segmentation based on transaction data is an important solution to group customers based on similar purchasing patterns. This study aims to examine customer segmentation based on sales transactions to help companies understand customer characteristics and develop more targeted and adaptive marketing strategies. A quantitative approach is used by applying the K-Means Clustering algorithm and PCA dimension reduction to a dataset from Kaggle containing 3,900 entries with 9 attributes. Determination of the optimal number of clusters was carried out using the Elbow and Silhouette Score methods. The segmentation results show five optimal clusters with the highest Silhouette Score of 0.81. Cluster 0 is the most dominant. PCA visualization shows a fairly clear cluster separation although there is little overlap. This study has succeeded in grouping customers based on purchase volume. Limitations of the study include the uneven distribution of clusters and it is recommended to add demographic attributes and evaluate other algorithms such as DBSCAN.]]>
Copyrights © 2025
