The study investigates the energy efficiency of photovoltaic (PV) panels installed on public transport buses. The focus is on assessing urban routes for their solar energy harvesting potential and the impact of generated solar energy on fuel consumption. The analysis is based on real-world data collected from the public transport system in the city of Lublin, Poland. Due to the prevalence of GPS signal loss, particularly in dense urban environments, a machine learning-based algorithm was developed to enhance bus trajectory reconstruction. The proposed methodology comprises four main stages: (1) filtering and processing GPS data to reconstruct bus routes, (2) augmenting the dataset by computing additional metrics and factors from the original dataset, (3) dividing bus routes into hexagonal spatial units and calculating average metric values for each region, and (4) identifying areas with higher shading factors and solar energy potential for each spatial unit. The reconstruction stage uses anomaly detection techniques to mitigate the effects of incomplete or noisy GPS signals. The results indicate that several areas in Lublin are suboptimal for installing PV panels, that energy generated by roof-mounted solar panels positively influences the vehicle’s energy balance, and that the issue raised in the article could help improve urban planning. Furthermore, the proposed approach facilitates the identification of spatial patterns affecting PV performance and supports data-driven decision-making for planning energy-efficient transport systems