Cities are the primary source of global energy consumption and greenhouse gas emissions. Therefore, innovative methods for sustainable energy planning are required in urban areas. The spatial complexity and heterogeneity of cities are frequently difficult for urban energy models to capture. By combining Geographically Weighted Regression (GWR) with geospatial K-means clustering, this study seeks to improve urban energy analysis ans effectively support policy making. We first combine census sections into homogeneous super-tracts using open-source geospatial and socio-economic data. Next, GWR is used to look into the relationships between important urban characteristics and residential electrical energy consumption. The methodology includes an iterative step to optimise the number of clusters, considering the statistical performance of both algorithms. In addition to identifying different urban typologies and critical factors influencing energy use, such as population density, green cover, the percentage of elderly people, and building characteristics, the proposed methodology highlights significant local variations in their impact. The results for Naples, Italy, show that this integrated approach has the potential to give a better understanding of urban energy patterns, which would strengthen the analytical basis for future AI-driven decision-support systems for policymakers and targeted energy efficiency interventions.