This paper reviews machine learning methods for solar radiation forecasting. The objective is to provide an overview of forecasting methods for solar irradiation using machine learning approaches. Various methods, including neural networks, support vector regression, regression trees, random forests, and gradient boosting, are discussed. The performance of these methods is complicated by the diversity of data sets, time steps, forecasting horizons, and performance indicators. Overall, the prediction error is quite equivalent. To improve prediction performance, hybrid models or ensemble forecast approaches are proposed. The paper also discusses the necessity of predicting solar radiation for effective operation of the power grid and optimal management of energy fluxes. The integration of renewable energy sources into the electrical grid increases the complexity of grid management due to the intermittent and unpredictable nature of solar radiation. The paper reviews various forecasting methodologies, including cloud imagery combined with physical models and machine learning models. It also discusses the use of machine learning methods for solar radiation forecasting, including classification and data preparation, supervised learning, unsupervised learning, and ensemble learning. The paper evaluates the accuracy of forecasting models using various metrics, including mean bias error, mean absolute error, mean square error, root mean square error, and mean absolute percentage error. The paper concludes that the accuracy of forecasting models depends on the location and time period used for evaluation, and that comparison of different forecasts against a common set of test data is necessary. The paper also discusses the use of reference models, such as the persistence model, for benchmarking. The paper highlights the importance of forecasting solar radiation for effective operation of the power grid and optimal management of energy fluxes.This paper reviews machine learning methods for solar radiation forecasting. The objective is to provide an overview of forecasting methods for solar irradiation using machine learning approaches. Various methods, including neural networks, support vector regression, regression trees, random forests, and gradient boosting, are discussed. The performance of these methods is complicated by the diversity of data sets, time steps, forecasting horizons, and performance indicators. Overall, the prediction error is quite equivalent. To improve prediction performance, hybrid models or ensemble forecast approaches are proposed. The paper also discusses the necessity of predicting solar radiation for effective operation of the power grid and optimal management of energy fluxes. The integration of renewable energy sources into the electrical grid increases the complexity of grid management due to the intermittent and unpredictable nature of solar radiation. The paper reviews various forecasting methodologies, including cloud imagery combined with physical models and machine learning models. It also discusses the use of machine learning methods for solar radiation forecasting, including classification and data preparation, supervised learning, unsupervised learning, and ensemble learning. The paper evaluates the accuracy of forecasting models using various metrics, including mean bias error, mean absolute error, mean square error, root mean square error, and mean absolute percentage error. The paper concludes that the accuracy of forecasting models depends on the location and time period used for evaluation, and that comparison of different forecasts against a common set of test data is necessary. The paper also discusses the use of reference models, such as the persistence model, for benchmarking. The paper highlights the importance of forecasting solar radiation for effective operation of the power grid and optimal management of energy fluxes.