Weather prediction using random forest machine learning model

Rajasekaran Meenal, Prawin Angel Michael, D. Pamela, Rajasekaran Ekambaram

Indonesian Journal of Electrical Engineering and Computer Science · 01 May 2021

In one sentence. Random forest predicts daily solar radiation and wind speed for Tamil Nadu from station data, reaching R2 0.97 and beating regression and SVM baselines.

Abstract

The complex numerical climate models pose a big challenge for scientists in weather predictions, especially for tropical system. This paper is focused on presenting the importance of weather prediction using machine learning (ML) technique. Recently many researchers recommended that the machine learning models can produce sensible weather predictions in spite of having no precise knowledge of atmospheric physics. In this work, global solar radiation (GSR) in MJ/m2/day and wind speed in m/s is predicted for Tamil Nadu, India using a random forest ML model. The random forest ML model is validated with measured wind and solar radiation data collected from IMD, Pune. The prediction results based on the random forest ML model are compared with statistical regression models and SVM ML model. Overall, random forest machine learning model has minimum error values of 0.750 MSE and R2 score of 0.97. Compared to regression models and SVM ML model, the prediction results of random forest ML model are more accurate. Thus, this study neglects the need for an expensive measuring instrument in all potential locations to acquire the solar radiation and wind speed data.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.