A Predicted Approach for Solar Radiation Using Multivariate Time Series

Sun's radiation (Rs) refers to the sunlight that arrives at the Grounds cape and provides light, heat, and energy. It plays a crucial role in sustaining life on our planet. Capturing and converting Rs into usable forms of energy, such as heat and electricity, is possible with different devices. However, measuring Rs accurately can be challenging due to various factors. Consequently, alternative methods have been developed to estimate Rs using different parameters and models. In this study, a model based on Multivariate Time Series (MVTS) delay values is introduced to examine how different feature selection (FS) approaches impact the prediction of Rs. The methodology proposed involves utilizing a random forest model to iteratively eliminate features, and subsequently evaluating its performance in comparison to Logistic Regression and Decision Tree models. The obtained results demonstrate that the approach produces reliable predictions based on several important criteria. Interestingly, the ranking of features varied depending on the model used, even when applied to the same dataset. The Logistic Regression model performs exceptionally well based on the Route Mean Square Error (RMSE) and coefficient of determination (R2) scores, although the other models also exhibit impressive performance.