Using K-Nearest Neighbor and Random Forest Approaches for Classifying Solar Radiation

Abstract

Studying climatic status and meteorological effects is important to identify climatic and environmental elements and their impacts in various fields of human life as well as other organisms. In this study, solar radiation (SLR) variables will be studied and classified based on their autoregressive variables by identifying the mathematical relationship among these variables using K-nearest Neighbor (KNN) and Random Forest (RF) techniques. Iraqi datasets taken from an agricultural meteorological station located in Mosul, Iraq, were used as a real case study. In these types of data, there are many obstacles, including nonlinearity and uncertainty, that will be the reasons for inaccurate classifications. The results of the comparisons explain that the RF approach and KNN in SLR classification have varied classification performance, while both of them produce highly accurate classification results. In conclusion, SLR can be accurately classified using RF and KNN techniques.