Application of machine learning for renewable energy prediction

Global horizontal irradiance or GHI is the amount of shortwave radiation acquired from above by a surface horizontal to the ground. It is of great significance in photovoltaic installation. GHI value is used to compute flat-panel PV output. In this paper we have discussed how to predict GHI value by using previous night's weather data and evaluated the predictions generated by various machine learning algorithm like Decision Tree Regression, Random Forest Regression and XGBoost Regression algorithms. Upon training we found that XGBoost Regressor was generating the best output of all the models we developed. We have evaluated the accuracy based on the metrics explained variance score.