An Efficient Supervised Machine Learning Model Approach for Forecasting of Renewable Energy to Tackle Climate Change

Abstract

This paper aims to introduce a reliable forecasting model for the consumption of electricity using renewable sources (namely: offshore wind, onshore wind and solar power) in EU countries, based on live data from the ENTSOE transparency platform as its input. The primary use behind this data science and machine learning methodology, is to help judge the availability of renewable energy resources. Aforementioned software is put to work by inputting desired country and associated parameters. It learns by carefully observing past patterns and their seasonality to make accurate predictions for the future. The ML algorithms used in this process are linear regression, extra trees regression, random forest regression, support vector machine (SVM) and gradient boosting, and precision is substantiated by getting a minimal Symmetric Mean Absolute Error (SMAPE) of 1-2.