Comparative Analysis of Machine Learning Models for Relative Humidity Prediction in the Philippines

Relative humidity is an important environmental parameter and is widely used in various fields. Prediction of humidity levels is crucial for climate modeling, heat stress, air quality forecasting, and public health. Machine learning techniques have shown potential for predicting humidity due to their nonlinear nature. However, there is a research gap in humidity prediction in the Philippines, specifically the lack of studies utilizing the available parameters provided by PAGASA, presenting an opportunity for further investigation and development of models for predicting humidity levels in the country. In this study, the researchers used a publicly available dataset from PAGASA containing weather measurements from 2000 to 2022 in the Philippines. Various machine learning models were trained and tested, with hyperparameter tuning performed using Bayesian optimization. The Gaussian Process Regression model with optimized hyperparameters achieved the best performance in predicting relative humidity, with the lowest RMSE and highest R-squared values. This study provides a reliable way to predict humidity levels in the Philippines based on weather parameters.