Forecasting Model of Electricity Consumption in Malaysia: A Geometric Brownian Motion Approach

Abstract

We learn from the literature that most of time series datasets in real world application consist of positive data.
However, no method deals specifically with this type of dataset. In this paper a model for such dataset using
geometric Brownian motion (GBM) is introduced and then employed in modelling electricity consumption in
Malaysia. We show that the model is not only better in terms of mean absolute percentage error but also in terms of
modelling running time. As far as we know, the use of GBM in developing time series model when data are positive
is unprecedented.