Statistical Analysis of Prosumers Connected to the Low-Voltage Distribution Network in the Banat Region, Romania

This paper presents a statistical analysis of prosumers connected to the low-voltage distribution network in the Banat region of Romania, based on an initial dataset comprising 21,457 records collected between 2019 and 2024. After removing incomplete or inconsistent records, the analysis was performed on a final sample of 21,426 observations. The methodology employed includes descriptive statistics, correlation analysis, simple and multiple linear regression models, as well as forecasting techniques based on simple and weighted moving averages. The results revealed a very strong correlation between installed capacity and exported power, confirming the relevance of these indicators for assessing the performance of photovoltaic systems. The multiple linear regression model, with exported power as the dependent variable and installed capacity, approved power, and active energy consumption as independent variables, achieved a high coefficient of determination (R²=0.953). The application of simple and weighted moving average methods enabled the identification of consumption and energy production trends over the analyzed period. Furthermore, the comparison of forecasting errors demonstrated that multiple linear regression provides the highest accuracy in estimating actual values, outperforming moving-average-based forecasting methods. The findings contribute to a better understanding of prosumer behavior within low-voltage distribution networks and may support the analysis, planning, and operation of regional energy systems.