This project investigated ensemble learning methods for day-ahead electricity operational forecasting across all National Electricity Market (NEM) regions in Australia, in collaboration with Australian Energy Market Operator (AEMO) and the Department of Industry, Science, Energy and Resources (DISER), Australia. Using historical operational demand and weather data, machine learning methods were developed that can select and combine from multiple forecasting models to maximize the forecasting accuracy.Prediction is obtained by combining the predictions of all individual models.In this work, the model diversity is obtained by using different part of the data. Specifically, half-hour data are modelled separately resulting in 48 prediction models used for the forecasting.