Solar Power Prediction using SARIMA, XGBoost and CNN-LSTM

The purpose of this post is to show how the application of data science methodologies can be used to solve problems within the renewable energy sector. I will discuss techniques to gain tangible value from a dataset by using hypothesis testing, feature engineering, time-series modelling methods and much more. I will also address issues such as data leakage and data preparation for different time series models and they can be managed. The energy sector has seen a rise in harnessing renewable energy to provide homes with electricity, however, whether it be on a large scale or for domestic use, the problems remain the same. Power plants which provide electricity sourced from renewable sources, face the difficulty of intermittency and need constant maintenance.