Statistical Modeling of Complaint Resolution Times in the Paraguayan Power System

Abstract

The National Electricity Administration (ANDE), responsible for Paraguay’s electricity system, developed the Complaint Management System (GRA) to improve the handling of reports of power outages. The aim of this work is to analyze the resolution times for complaints about power outages by statistically processing historical data from the GRA between May 2021 and June 2024. To this end, a rigorous cleansing of a database of 2,875,287 records was carried out, resulting in a final set of 270,209 valid complaints. The study of this data showed that the distribution of resolution times was positively skewed, with a median of 1.79 hours and a mean of 2.74 hours. Sixty-five point seven percent of the total time was concentrated in administrative stages, which showed higher Spearman correlation coefficients than field tasks. We also identified weak correlations (p ≤ 0,11) between the number of claims and variables such as the minimum daily temperature. Kruskal-Wallis tests revealed significant differences in claim times and volumes according to geographical and temporal variables. Likewise, the KolmogorovSmirnov test indicated that the Log-Normal distribution offers the best fit (D = 0,007) for modeling resolution times. The results highlight the influence of geographical, temporal, and technical factors on resolution times, which helps to distinguish different patterns in the system’s operation, suggesting improvements through automation, training, workflow redesign, and efficient resource management