Clustering Analysis of PLN XYZ Customers Using K-Means and Gaussian Mixture Model (GMM) for Marketing Strategy

The clustering method is useful for companies to identify hidden information such as types of customer groups and demographics based on their personal information, such as gender, age, salary, and transaction patterns. Clustering can help managers make important business strategies and decisions. In this study, transaction data from electricity purchases by customers of Perusahaan Listrik Negara (PLN) XYZ is analyzed. The purpose of the customer segmentation process is to develop targeted marketing strategies for each group, ensuring that marketing efforts are more effective once customers are grouped into clusters. This research creates clusters from transaction data by comparing two clustering methods, k-means and GMM (Gaussian Mixture Model), by transforming the attributes into LRFM values: the duration since the customer’s first transaction (L), the recency of their most recent transaction (R), the frequency of their transactions (F), and the total transaction value within a defined period (M). In the cluster analysis research using PLN transaction data with four clusters, the k-means method received a silhouette score of 0.63, which was higher than the GMM method’s score of 0.17.