Meteorological Data Analysis and Precipitation Prediction Using Machine Learning: A Case Study of East Asia

Accurate precipitation forecasting is crucial for various industries, including agriculture, disastermanagement, and urban planning. The current study applies the ERA5 dataset for the prediction of totalprecipitation in East Asia using machine learning techniques with a focus on the novel integration of laggedtemperature and wind speed variables with the Linear Support Vector Regression (LinearSVR) model. Incontrast to earlier research that mostly employed sophisticated models such as Random Forests or deeplearning, our strategy focuses on the limitations of employing basic linear techniques in capturing intricatemeteorological relationships. The performance of the model, measured by Mean Squared Error (MSE) andR-squared (R²) metrics, reflects its inability to capture non-linear dynamics, leading to poor predictions.Some of the main challenges are the absence of non-linear feature interactions and few input variables.Follow-up research will investigate using additional meteorological parameters and using higher-levelmodels, i.e., ensemble methods or artificial neural networks, to improve predictive ability.