Classification of Dengue using Machine Learning Techniques

Abstract

Dengue infection belongs to the family of virus, Flaviviridae, consisting of four serotypes which spread through the chomp of contaminated Aedes mosquitoes. Around 2.5 billion individuals live in dengue-hazard locales with around 100 million new cases every year around the world. The worldwide predominance of dengue has grown dramatically in later decades. The illness is now endemic in more than 100 nations in Africa, the Americas, the eastern Mediterranean, South East Asia and the western pacific south Asia and the western pacific are the most genuinely influenced. In 1970's only nine nations had encountered DHF plagues, a number which had expanded more than four- crease by 1995[30].Numerous clinical signs are utilized for diagnosing of fever. In any case, it has been an awesome test for the doctors to distinguish the level of hazard in dengue patients utilizing clinical indications. But the disadvantages of clinical procedures make machine learning more powerful in diagnosing of fever in affected patients. Subsequently, this study plans to apply a non-invasive machine learning techniques to help the doctors for ordering the hazard in dengue patients. Conducted a comparison study among Simple Classification and Regression Tree(CART), Multi-layer perception (MLP) and C4.5 algorithms, based on which demonstrating that Simple CART algorithm shows 100% accuracy for classification of affected or unaffected patient.