Atmospheric Temperature Prediction Using SVR Models

The weather is the state of atmosphere that can be described with various categorical and quantitative parameters. A lot of everyday operations (ex: logistics) depend on these parameters and meteorologists all over the world are trying to predict them as accurately as they can. In this paper, one atmospheric temperature prediction method is proposed using Support Vector Regressor (SVR) Machine Learning model. This study goes through the preprocessing stages and features selection based on a realworld dataset, determining which features are predictable by SVR, proposed hyperparameters and other adjustments. Extensive tests were conducted which demonstrated the proposed solution's potential: the algorithm produces reasonable 1-day predictions with an acceptable margin of error considering the volatility of the matter.