European Journal of Molecular & Clinical Medicine

This paper presents a study on minimising the estimation error of forecasting the
electricity consumption in Malaysia. A robust and accurate forecasts of electricity
consumption are deemed crucial for the supplier to arrive on fair estimations of electricity
supply optimally. Thus, identifying the best model to forecast the electricity consumption
accurately may hinder energy wastage. This research aims to examine which model gives
the least error in estimating the future electricity consumptions in Malaysia. Two models
were tested namely Artificial Neural Network (ANN) and Regression Methods. In analysing
these models, this research applies the Microsoft Excel and SAS Enterprise Miner (SAS)
software. The data were extracted from the Department of Statistics Malaysia (DOSM),
CEIC Data Company and The Statistics Portal. Results indicate that ANN produces least
error as compared to the Regression Method as the former fits the data well whilst the latter
overfits the data. The ANN model uses NNTool from MATLAB is used for forecasting
future electricity consumption. The forecasted values (2020-2022) proved to provide more
interpretable forecasts. This study may benefit the electricity supplier, consumers and also
the Government of Malaysia, in particular the Ministry of Energy and Natural Resources.
It may provide insights on estimating the optimum amount of energy to be generated. This
will definitely increase the savings and reduce wastage from every angle. Ultimately, the
environment is saved too.