Enhancing predictive accuracy in environmental data analysis: a hybrid LASSO-RFR approach for climatic analysis in Siberia

Abstract

This study introduces a hybrid LASSO-RFR approach for photovoltaic energy forecasting, leveraging LASSO's feature selection with RFR's analytical strength to tackle weather-induced variability. It showcases improved forecast accuracy through simplified datasets and enhanced correlation analysis, resulting in superior model performance. With an MSE of 0.0060 and an R squared of 85.7% for Model 2, the approach outperforms LASSO-only models, marking a significant advancement in renewable energy analytics and offering a potent forecasting tool for areas with extreme weather.