Analyzing and forecasting covid-19 outbreak in India

Abstract

The unprecedented outbreak of the COVID-19 virus has infected more than 50 million people all over the world in less than a year. More than 1 million people have lost their lives due to the ongoing pandemic. The pandemic struck India on January 30, 2020, when the first positive case of COVID-19 was identified in Kerala. Today, India is one of the most adversely affected countries in the world. Hence, it is of utmost importance to analyze the trends in India and use the adopted knowledge to forecast the future course of outcomes. Along with the overall trend analysis in India, this study also takes into account 5 most affected states of the country: Maharashtra, Andhra Pradesh, Tamil Nadu, Karnataka and Uttar Pradesh as the subjects of the research. ARIMA and Prophet time series forecasting models have been used to make three types of predictions: confirmed cases, deaths and recovered cases in India as well as in the adopted states. The effectiveness of the forecasting models is evaluated based on metrics such as Root Mean Squared Error, Mean Absolute Error, Mean Absolute Percentage Error and Coefficient of Determination. The results suggest that the adopted models are promising mechanisms for forecasting COVID-19 trends. Our study also suggests that ARIMA model performs better than Prophet Model at this task of forecasting the outbreak. The forecasts can be useful in increasing the preparedness level of government authorities, health facilities and hospitals to combat against massive spread of the virus.