GRU-Based Forecasting Model for Energy Production and Consumption: Leveraging Random Forest Feature Importance

There is huge ignorance of the energy situation in the world. People in different countries are unaware of the amount of consumed energy and how this accelerates the depletion of non-renewable energy resources which might lead to unpredictable consequences. Moreover, the continuous increase in the amount of produced and consumed energy increases the amount of CO2 emission which leads to series pollution in the air. These problems can be solved initially by increasing awareness of the general situation in every country and seeking to solve these problems. In this project, we aimed to develop prediction models for energy production, consumption, CO2 emissions, and share of renewable energy in electricity production. We collected data from several resources and used the random forest feature importance method to select features with high importance scores to produce highly accurate prediction results. We implemented the models using a Gated Recurrent Unit neural network which generated very highly accurate results because of its ability to remember information from the past. In the future, we seek to implement prediction models for each country to increase awareness of energy-related subjects.