REVIEW ON VARIOUS FORECASTING METHODS ON RENEWABLE ENERGY SOURCES

New issues such as grid convergence, load balance and trading have been brought on by a growing presence of renewable energy plants, which makes successful prediction models essential. Recent literature approaches have shown that taking advantage of spatio-temporal autocorrelation in data derived from many plants will lead to better predictions. Though tensor models and techniques for handling spacetemporary data are appropriate, little attention has been paid to them in the field of electricity. In this paper we provide a new approach based on the decomposition of the Tucker tensor which can extract a new functional space for learning. In comparison with the original function space, we examined the efficiency of the predictive cluster trees in three renewable energy data sets. The findings are positive even in comparison with state-of-the-art algorithms for the approach proposed.