A Predictive Analysis of Residential Electrical Load Demand in Mauritius

Electrical Energy is one of the most common forms of energy in the world. Due to the increasing socio-economical style of living, the demand for electrical energy is increasing at an unprecedented rate, which transducers to high emission of greenhouse gases due to the combustion of fossil fuel. Two approaches can be applied to mitigate the emission; increased use of renewable sources of energy or retrofitting existing systems with technology. The latter was explored in this study. Predictive analysis was employed to predict the load demand for a typical Mauritian household. Fuzzy Time series and Probabilistic Neural Network were considered as predictive models and were modelled in KNIME Platform Analytics. It was found that Probabilistic Neural Network was more appropriate for the load demand prediction yielding a Mean Absolute Percentage Error of 3.5 % as compared to the 4.7 % for Fuzzy Time series.