Performance Comparison of Simple Regression, Random Forest and XGBoost Algorithms for Forecasting Electricity Demand
Muhammet Mustafa Gökçe, Erkan Duman
Abstract
Electrical energy is the locomotive of the economy, industry, and development in terms of the development of countries. In order to meet the need during the periods when the energy demand reaches its peak and to prevent the market participants from making economic losses when it is at the lowest level, the closest prediction should be made. Load forecasting is very important in planning the generation, transmission, and management of energy and in pricing electricity in the most appropriate way. Regional, demographic and meteorological variables are effective in the energy production plan. These factors affect the electricity market operated by the system operator in every sense. An energy forecasting plan is needed in order to keep the supply and demand of energy in balance. Today, the use of large data sets has a positive effect on machine learning and artificial neural network training. By using these data sets, very high performances are achieved in modeling. In this study, Turkey's electricity consumption between the years 2018-2021 was modeled using the Linear Regression from supervised learning techniques, Random Forest and XGBoost algorithms from machine learning models. In our study, short-term consumption load forecastings were made hourly, considering the meteorological factors and public holidays in the country, and the forecasting performances of the three different algorithms used were compared. If the study is used, it is foreseen that it will eliminate the energy supply-demand imbalance.