Improving forecasting through a hybrid Theta method and its integration with multiple temporal aggregation