Electricity Demand Forecasting in Buildings Based on ARIMA and ARX Models

The Accuracy of electricity demand forecasting is a key success factor of the organizational operation since energy is the crucial driven force of all activities. As a result, if executives in any organizations can accurately predict the future demand of the electricity consumption. They will be able to plan ahead the budget regarding the electricity bill as well as the energy conversation initiatives of the organization. Another advantage is the capability to estimate the impact of electricity usage on the environment since the generation of electricity always leads to the consumption of natural resource, e.g., water, and the release of greenhouse gas to the atmosphere. Due to the study, the electricity demand from January 2015 to November 2018 of seven faculty buildings in a University was monthly recorded. Autoregressive Integrated Moving Average (ARIMA) model was utilized to model the time series demand. Since there is another information regarding the number of students from the same period, another forecasting model, Autoregressive with Exogenous Output (ARX), was also used. The results in term of forecasting error show that the ARX outperforms the ARIMA model especially when the lagging order of ARX is high.