Метод автоматической генерации признакового пространства в задаче прогнозирования потребления электроэнергии

Информационные технологии. 2021. Т. 27. № 10. С. 550–560.

Danilov K., Maltseva S. V.

The automated feature engineering method in the problem of forecasting energy consumption is considered. The algorithm of the method and the scheme of the forecasting model construction are stated. The proposed approach was tested on data about electricity consumption in Russian regions. The results of the computational experiments carried out using the described method demonstrate an increase in the efficiency of the developed forecasting model and improvement of accuracy.

Keywords: автоматическое машинное обучение, bayesian optimization, automated machine learning, automated feature engineering, forecasting energy consumption, автоматическая генерация признаков, прогноз потребления электроэнергии, байесовская оптимизация