Using the standardised SORT, the article analyses instantaneous energy consumption and recuperation processes in an electric bus. The test includes three scenarios: SORT 1 (heavy urban traffic), SORT 2 (mixed driving conditions), and SORT 3 (suburban routes), enabling precise assessment of the energy efficiency of vehicles while eliminating environmental variables. The recuperation system significantly enhances energy efficiency, though its effectiveness varies based on the driving scenario. Modelling methods were compared as follows: linear regression, KNN algorithms, and neural networks, achieving a high fit (R2 > 90%). While KNN and neural networks were better at reproducing nonlinearities, they indicated the need for additional variables and time delays to enhance accuracy. The article sets itself apart by incorporating predictive models and examining recuperation efficiency across various scenarios. It emphasizes the importance of combining SORT results with real operational data and developing adaptive energy management systems. The results indicate the potential for optimizing electric buses for public transport, including route planning and further improving recuperation technology, which can significantly reduce energy consumption and greenhouse gas emissions.