System approach to forecasting electricity demand based on machine learning

The article explores a systemic approach to solving the problem of forecasting electric-ity demand in Ukraine based on machine learning methods. The sequence of data processing stages when solving the forecasting problem using machine learning methods is presented. The methodology for solving the forecasting problem on time series is considered. The fore-casting process consists of five stages. The first stage includes data collection, analysis and interpretation. The next stage includes data research and preparation procedures. The third stage - the modeling stage consists of three parts: preparation of a data set for modeling, se-lection and training of models and assessment of their quality. The fourth stage is the fore-casting stage and calculation of quality indicators of forecasts. At the fifth stage, procedures are performed to increase the efficiency of the selected forecasting model. At the modeling stage, the following models were used: ARIMA, GAM, ANN and BSTS. The models were ana-lyzed and forecasts were built based on each model. For the built models with the best quality indicators, forecast values were calculated. The forecasts were compared with the data of the validation sample. To select the optimal model, the following indicators were used: MAPE, MAE, MSE, RMSE. The BSTS model showed the best results.