Power consumption forecasting model utilizing ensemble learning of LSTM networks in the development of a meta-model which explores the similarity and causality of clients and aggretates the results through stacked generalization in an MLP network.