Comparative analysis of various supervised machine learning techniques for diagnosis of COVID-19

Coronavirus is a large family of a viruses that causes illness ranging from a normal cold to severe disease. COVID-19 is another strain that has not been distinguished in humans before. As this virus is rapidly spreading all over the globe, we need to implement a mathematical model to estimate the prediction of new cases as well as how to classify that a person is COVID-19 positive or not by considering the practical scenario in India. In this research, we proposed three different supervised machine learning techniques for diagnosis of COVID-19. We have compared classification results of different techniques, i.e., bagging algorithm, k-nearest neighbor, and random forest for classifying the datasets of COVID-19. For the classification purpose, we took symptoms from a Covid-19 tracker in India, whereas India has entered into the second stage. The performance of each technique is evaluated using various performance measures. The classification results show that the random forest gives better results, employing accuracy of 85.71% and F1 score of 0.833.