Analysis of BKKBN Service Satisfaction Data Segmentation Model Using Hybrid PCA and K-Means Approach to Optimize Customer Experience Management

Abstract

This study aims to develop a model for segmenting public service satisfaction at the East Java BKKBN using a hybrid approach that combines Principal Component Analysis (PCA) and K-Means Clustering to optimise Customer Experience Management (CEM). The data used came from 13 service quality indicators and was analysed descriptively, showing a good level of respondent satisfaction (mean 3.72–3.84). The factor suitability test produced a KMO value of 0.969 and a significant Bartlett's Test (p-value 0.00), indicating that the data was suitable for factor analysis. PCA was used to reduce the data dimensions and identify the main components that explained 55% of the total variation. The reduction results were then grouped using the K-Means method and produced eight optimal clusters (Silhouette 0.532; Davies-Bouldin 1.59; Calinski-Harabasz 370.60). Only one cluster (29.4% of respondents) showed a very high level of satisfaction, while the others were classified as moderate to low. These results confirm the need to improve the efficiency, responsiveness, and transparency of services. The hybrid PCA–K-Means approach proved effective in revealing patterns of public satisfaction and became the basis for data-driven policy-making in improving the quality of public services