Managing energy has emerged as a significant concern for digital enterprises due to their environmental impact. The approaches employed within the sphere of incorporating energy into databases consist mainly in building an energy cost model for a load of queries in order to predict the energy consumption of the system, then applying an cost-driven technique focused on energy optimization by revisisting query processing, algorithm planning, and update techniques. Creating an energy cost model involves several distinct stages, starting with a system audit to pinpoint crucial parameters and progressing to the calculation of energy quantities through the utilization of a machine learning (ML) methodology. Machine learning has rapidly emerged as a technology widely employed across various industries and stands as one of the most rapidly advancing domains. It encompasses a multitude of algorithms aimed at accomplishing common tasks such as prediction and classification. In this study, we evaluate three predictive techniques in formulating an energy cost model for an analytical query. The primary goal of this paper is to present a universal approach for constructing energy models during query processing in a relational database and demonstrate the influence of the machine learning phase on the final model’s accuracy.