Development of energy consumption prediction models based on gradient descent methods

The purpose of this paper is to develop models for predicting energy consumption of buildings based on gradient descent methods, analyze their effectiveness and compare them with other machine learning methods. To achieve this purpose, the following tasks were performed: analysis of gradient descent and its modifications Momentum, Heavy Ball and Nesterov, data preprocessing, realization of prediction models based on gradient descent methods and evaluation of model performance by MSE, RMSE and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> metrics. Additionally, other machine learning methods, such as Decision Tree, Random Forest and K-Nearest Neighbors, were used for comparison. The results showed that the Heavy Ball method provides the best accuracy among all the approaches considered, and the gradient descent methods generally demonstrated higher efficiency compared to the other methods. The novelty of this paper is to study the effectiveness of gradient methods and compare them with other machine learning methods for energy consumption prediction. The obtained results confirm the effectiveness of gradient methods in regression tasks for energy consumption prediction and emphasize their importance in modern machine learning, in particular in the energy sector.