Forecasting Energy Production from Coal, Gas and Coal Sources in Pakistan by Using Machine Learning Models

Abstract

For energy planning and policymaking, forecasting energy production is significant. Pakistan's energy production through oil, gas, and coal sources over the next ten years was predicted using data from 1971 to 2015. The forecasting procedure is carried out with the help of the autoregressive integrated moving average (ARIMA) model, Holt-Winter Exponential Smoothing model, Artificial Neural Networks (ANNs), and Hybrid Model. The ARIMA (1,1,0) and NN (2,2) are determined for the data. The annual electricity production from oil, gas, and coal sources will be 58% to 65% (of the total production) in the next ten years. This research holds enormous significance for Pakistan's energy landscape. Dependable and exact energy creation conjectures are fundamental for policymakers, energy organizers, and partners in making informed choices on asset designation and infrastructure development. By guaranteeing the manageability, dependability, and productivity of Pakistan's energy framework, these forecasts assume an imperative part in supporting economic growth and satisfying the population's energy needs. Furthermore, the study investigates a scope of modeling methods, including ARIMA, Holt-Winter Exponential Smoothing, and ANNs, and presents a Hybrid model. The hybrid approach, which combines the qualities of different models, offers a promising answer for upgrading forecasting accuracy. By considering the complex elements of electricity production from various sources, this exploration benefits Pakistan and adds to the more extensive energy forecasting field.