Forecasting epidemic spread of COVID-19 in India using arima model and effectiveness of lockdown

Abstract

In the present situation, coronavirus which renamed as COVID 19 by World Health Organization is the most infectious illness spreading worldwide whose first case saw in China on December 31, 2019. India is the second most populated nation in the world; accordingly, it is a state of research to forecast the COVID-19 spread in India. Till May 03, 2020, 42,505 COVID-19 cases were confirmed in India including 1,391 deaths. COVID-19 infected patient data has extracted from the covidindia, kaggle and World Health Organization website which includes daily confirmed, recovered and death cases from February 15, 2020 to May, 03, 2020. In the current study, the data-driven estimation model ARIMA is utilized for determining the number of COVID-19 cases in India 76 days ahead from 03 May, 2020 using R statistical package. Predictions were done with average of 93.695% of accuracy for confirmed case models, 86.96% of accuracy for recovered case models, 87.94% for death case models and 90.91% for death rate respectively. The 76-day forecasting of COVID-19 contaminated patients could be ascend to 28,2529, recovered cases could be expanded up to 11,9046 and death cases could expanded up to 9675 at July, 18 2020 with Confidence Interval of 80% to 95%. This pioneering study mainly focus on the importance of nationwide three lockdown, self isolation and social distancing in control the disease transmissibility among Indian population through data driven ARIMA model analysis. The ARIMA model with Akaike’s Information Criterion, Bayesian Information Criterion, and Ljung-Box Statistics test utilized for best model determination which can quickly help in estimating COVID-19 contaminated patients and help administration to make better decision to fight with this contagious disease.