Understanding Solar Power: Analyzing and Predicting Photovoltaic Energy Output

Abstract

This paper compares data-driven  algorithms (Linear Regression, Random Forest, and  Decision Tree) for solar energy prediction. It analyzes  variables like Daily Yield, Total Yield, Ambient  Temperature, Module Temperature, Irradiation, and DC  Power using a dataset with unprecedented granularity. The  algorithms were trained and tuned for optimal  performance, resulting in high accuracy levels. Linear  Regression achieved 99.4% accuracy, Random Forest  achieved 99.2% accuracy, and Decision Tree had the  highest accuracy at 99.8%. The analysis identified  strengths and weaknesses of each algorithm, indicating  their suitability for different prediction scenarios. These  findings have significant implications for integrating solar  energy into the power system, instilling confidence in the  reliability of data-driven algorithms for precise solar  energy forecasting.