Decision Support System for Forecasting the trends in Energy Sector of Pakistan: Multivariate Modeling with Socio-economic and Environmental Factors

Abstract

Energy plays an important role in the economic growth of any nation, but unfortunately, Pakistan is facing a severe power crisis that has caused economic losses in the past two decades. One reason is the lack of proper planning, and planning is made with the help of accurate forecasting. Usually, energy consumption prediction is based on the univariate models neglecting the other influencing factors like economic, social and environmental. And even when we consider these external factors, it is very hard to select the appropriate factors. We have taken into consideration nine different external factors and also the lag values of dependent variables (Electricity, gas and oil). Upon making different combinations, we derived over 4000 regression models and our algorithms selected the best models based on the value of mean absolute percentage error and root-mean-square error. Our method provides a comprehensive approach for variable selection and best model selection for prediction.

Keywords: Multivariate Model, Social Economic Environmental Factors, Energy forecasting, Decision Support System