Assessing the Efficacy of Machine Learning Techniques in Solar Irradiance Prediction

Background: Researchers have utilized long short-term memory (LSTM) to store knowledge of learning brief interconnections in various applications, such as predicting solar irradiance, translation software, and speech recognition. In this work, the researchers focus on using time-series solar irradiance-based data for prediction. Methods: The researchers employ the LSTM model to forecast solar irradiance data. Additionally, they use the regression model for random forests to make predictions. Results: In this study, each feature is trained using both LSTM and Random Forests. The researchers evaluate the accuracy of the targeted features using fundamental metrics such as the root-mean-square error (RMSE) and Mean Absolute Error (MAE) for LSTM and Random Forests. Other metrics, such as the cross-validated accuracy score and the r2 score, are only employed for random forest to further understand the accuracy of the targeted features: Concluding Remarks: The researchers have successfully utilized LSTM and Random Forests to forecast solar irradiance data. By comparing the accuracy metrics of both models, they can assess the performance of the targeted features.