Data analysis of grid-connected solar setup and regression based predictive models

The paper presents analysis carried on a 3.6 KW grid-connected solar facility. Exploratory data analysis has been carried out using python libraries for the solar energy generation, load, energy exported and imported from the grid. The paper develops a SVM model and XGBoost model in order to predict energy exported back to the grid at the CARET facility in AMU. SVM and XGBoost are powerful machine learning tools useful for time series prediction. The paper further compares the mean squared errors of the two models and explores which model gives a better prediction.