Prediction of growth in COVID-19 Cases in India based on Machine Learning Techniques

Abstract

One of the biggest health challenges that the world has faced in recent times is the pandemic due to coronavirus disease known as SARS-CoV-2, or Covid-19 as officially named by the World Health Organization (WHO). To plan medical facilities in a certain location in order to combat the disease in near future, public health policy makers expect reliable prediction of the number of Covid-19 positive cases in that location. The requirement of reliable prediction gives rise to the need for studying growth in the number of Covid-19 positive cases in the past and predicting the growth in the number in near future. In this study, the growth in the number of Covid-19 positive cases have been modelled using several machine learning based regression techniques viz., Multiple Linear Regression, Decision Tree Regression and Support Vector Regression. Further, different feature selection techniques based on Filter and Wrapper methods have been applied to select the suitable features based on which prediction is to be done. This study proposes the best observed method for modelling the pattern of growth in number of Covid-19 cases in the near future for a locality and also the best selection method that can be employed for obtaining the optimal feature set. It has been observed that unregularized Multiple Linear regression model yields promising results on the test data set, compared to the other regression models, for predicting the future number of Covid-19 cases and Backward Elimination feature selection method performs better than other feature selection methods.