Predictive Model for Electricity Consumption in Malaysia Using Support Vector Regression

Electricity consumption is a significant indicator of modern society's development and advancement. It is influenced by factors such as population growth, urbanisation, and economic activity. However, predicting electricity consumption is a tough task due to the complexity and fluctuations of the energy market. In this paper, Support Vector Regression (SVR) was proposed in developing a predictive model for Malaysian electricity consumption. SVR was chosen as our proposed method as it can handle nonlinear and high-dimensional data using kernel functions. Data used for this study were retrieved from various sources including macrotrends.net, the World Bank's Climate Knowledge Portal, and the World Bank's indicator database. The dataset consists of relevant variables such as temperature, population density, and economic growth to anticipate future electricity demand. Results from this study showed that the SVR model outperforms other methods in terms of accuracy and error metrics. Additional components, hyperparameter fine-tuning, ensemble approach research, and long-term forecasting are all advocated for further improvement.