Comparison of machine learning models for the estimation of solar power generation in Australia

Abstract

Solar power is a renewable energy source that contributes a lot to improving the environmental sustainability of energy production. In Australia, rooftop solar panel systems are installed for power generation. By seeking the optimal installation strategy, people could develop more efficient technologies and systems. In this work, data related to power generation was gathered between January 1 and December 31, 2022. This article only analyzes qualitative data in the dataset. Firstly, the support vector machine (SVM) model is used to compare the results obtained by removing only one qualitative feature at a time with the results obtained by including all features. It is concluded that the shading condition feature has a significant impact on power generation, and then. By comparing the accuracy of K-nearest neighbor (KNN), Logistic Regression, SVM with Radial Basis Function (RBF) kernel, and Decision Tree, it was found that SVM with RBF kernel is the most suitable classification method for the feature of the shading condition.