Improving the Accuracy of Time Series Forecasting with Global Modelling and Ensembling

Accurate time series forecasting is highly beneficial for many types of businesses and industries, especially for improved decision making and strategic planning. Some industries had experienced huge financial losses due to the inaccurate forecasts provided by their forecasting models. This Ph.D. research aimed to address some existing issues in the forecasting space by using global modelling and ensembling which combines the forecasts of multiple models in the prediction process. Consequently, a set of novel and highly accurate forecasting frameworks are introduced where practitioners can incorporate them with their applications to obtain more accurate forecasts leading to high commercial benefits.