Predicting Dengue Fever via Weather Data: A Comparative Analysis of Statistical and Machine Learning Models

This study, based on meteorological factors, provides a forecast analysis of dengue cases in the Philippines. We used two datasets: one with the number of dengue cases and another with information on the weather such as surface pressure, mean tem- perature, relative humidity, wind speed, precipitation.Multiple machine learning models were utilized to predict future dengue cases: Random Forest classifier, CNN-1D, CNN with an attention mechanism, and Support Vector Regression (SVR). On the DS2 dataset, the CNN with an attention mechanism outperformed the other models, with a Root Mean Square Error (RMSE) of 0.02. Mean temperature is the most important meteorological param- eter impacting the rise in dengue incidence, according to feature importance analysis using SVR. The usefulness of sophisticated deep learning approaches in modeling epidemiological data is shown by this work, which also emphasizes the important role of weather parameters in dengue prediction.This paper highlights the significance of weather factors for dengue prediction and demonstrates the utility of advanced deep-learning algorithms in modelling epidemiological data.