Clustering of PLN ULP Binjai Timur Customer Complaints using the K-Means Method

Abstract

The large number of customer complaints received daily by PLN ULP Binjai Timur presents a challenge in providing responsive and accurate service. Irregularities in recording and grouping complaints mean that the available information is less than optimal for supporting decision-making. This study aims to group customer complaints based on similar characteristics for easier analysis. The method used is the K-Means algorithm, a clustering technique in data mining that divides data into several groups based on their proximity to the cluster center (centroid). The analysis was conducted through the Knowledge Discovery in Database (KDD) stages, which include data selection, transformation, and algorithm implementation using MATLAB software. The three main variables used in the grouping process were complaint type, complaint submission medium, and customer address. The implementation results in three main complaint clusters with distinct patterns, providing PLN with insight into the most frequently encountered problems, areas with high complaint rates, and the most frequently used reporting medium. These findings provide an important foundation for PLN in setting treatment priorities, improving service quality, and strengthening customer relationships. The application of the K-Means algorithm has proven effective as a systematic and practical solution for managing complex and large amounts of complaint data.