Evaluation of Machine Learning Approaches for Prediction of Dengue Fever

Abstract

Dengue is a mosquito-borne, deadly viral disease that is a major threat to public health all over the world. Dengue and covid-19 symptoms are almost same, and sometimes, people are confused about which disease they are infected with. This year in Bangladesh dengue and covid-19 patients have been increasing at an alarming rate, and most of the time people didn’t properly recognize the disease. A developing country like Bangladesh has faced many difficulties to handle this situation. The target of this research work is to analyze the symptoms and predict the chances to get infected with dengue fever. Machine learning techniques are widely utilized in the health industry to detect fraud in treatment at lower cost, predictive analysis, cure the disease. Four machine learning algorithms are used which are support vector machine, decision tree, K-nearest neighbor, random forest to predict dengue fever based on symptoms. The results were compared for percentage split and K-fold cross-validation method for before and after applying principal component analysis. The experimental result shows that the support vector machine algorithm provides the highest performance compared to others algorithms.