Medium-term load forecasting in isolated power systems based on ensemble machine learning models

Over the past decades, power companies have been implementing load forecasting to determine trends in electric system (EPS); therefore, is applied solve problems of management and development systems. This paper considers issue building a model medium-term graphs for EPS with specific properties, based on use ensemble machine learning methods. implements approach identification most significant features apply models an isolated system. A comparative study following was carried out: linear regression, support vector regression (SVR), decision tree random forest (Random Forest), gradient boosting over trees (XGBoost), adaptive (AdaBoost), AdaBoost regression. Isolation from time series allows implementation simpler more overfitting-resistant models. All above makes it possible increase reliability forecasts expand information technologies planning, management, operation EPSs. Calculations total forecast error proved that characteristics proposed are high quality accurate, thus they can be used real