Predicting Dengue Outbreak from Climate Data Using Machine Learning Algorithms

Abstract

Dengue is an Aedes mosquito-borne arboviral disease transmitted by the bite of a female Aedes aegyti and Aedes albopictus mosquitos in Bangladesh. The Directorate General of Health Services (DGHS) of Bangladesh reported that 28,429 cases have occurred from 1st January to 31th December in 2021 in Bangladesh. Many studies have consented that the emergence and incidence of disease of dengue is strongly correlated with meteorological and meteorological factors, particularly, temperature, rainfall and humidity. For that reason, during monsoon season (from June to September) dengue peak transmission occurs in Bangladesh. In this paper, we predict the accuracy of dengue outbreak from climate data. A dengue dataset, containing information of climate variables, dengue cases during 2019 to 2021 from Meteorology Department and Directorate General of Health Services (DGHS), Bangladesh. We split the whole dataset into 70:30 ratios were 70% considered as training and 30% for testing purposes. Such, prediction of accuracy we apply various supervised machine learning (ML) algorithms like Support Vector Machine (SVM), Decision Tree (DT), Logistic Regression (LR), Naïve Bayes (NB), AdaBoostClassifier (AdaBoost), XGBRegressor, GradientBoostingClassifier and Random Forest (RF). Finally, from these algorithms, SVM provide the highest accuracy of 96.73%.