POLA KEPUASAN PELANGGAN PDAM MENGGUNAKAN DATA MINING PADA BIG DATA OPERASIONAL DAN LAYANAN

Authors

  • Nanang Fahrurozi Sekolah Tinggi Teknik Ronggolawe
  • Joko Handoyo

Abstract

This study addresses the challenges faced by PDAM (Regional Water Company) in holistically understanding customer satisfaction due to fragmented data. By applying data mining to the integrated big data of Perumda Air Minum Tirta Amerta Blora, the study combined 7,766 data points from technical reports, meter calibrations, SIMPel complaints, the Si Tampan application, and social media using the CRISP-DM methodology. K-Means Clustering analysis identified four customer segments: Loyal Satisfied (18%), Technical Complaints (32%), Administrative Issues (29%), and Dissatisfied (21%). Three areas—Cepu, Ngawen, and Randublatung—were confirmed as critical areas with a Silhouette Score of 0.58. Random Forest Classification revealed five dominant satisfaction factors: complaint response time (0.224), water pressure (0.198), frequency of interruptions (0.156), meter status (0.142), and arrears (0.134). A strong correlation was found between water pressure and complaint frequency (r = -0.72). The predictive model demonstrated excellent performance with 89% accuracy, 88% precision, 85% recall, F1-score 0.86, and AUC-ROC 0.93. TF-IDF sentiment analysis confirmed the dominance of technical complaints on the terms "leak" and "water not coming out," as well as appreciation for digital innovation with 54% positive sentiment on social media. These findings recommend prioritizing infrastructure improvements in critical areas, optimizing complaint response times to less than 8 hours, developing a real-time analytics dashboard, strengthening Si Tampan's digital services, and a proactive social media strategy with rapid responses to continuously improve customer satisfaction.

Published

2026-06-30