Statistical Modeling of Covid 19 Outbreak in India: A Comparative and Predictive Analysis of Various Modeling Strategies

This study was aimed to apply various statistical models to predict the current and future cases of novel coronavirus (COVID-19) in India to highlight the best fit model. We utilized the secondary data of cumulative confirmed and recovered cases, and deaths due to COVID-19 that was available in public domain since 30th January 2020. Various statistical models, including exponential model, polynomial and time series (ARIMA) model, were utilized to predict the current cases of COVID-19 in India and model fitness assessed. The exponential model gave satisfactory R-squared value with a growth rate of 1.14. However, there was less-than-perfect fit of the predicted and available values by exponential model. The polynomial model and ARIMA model both gave R-squared value better than the exponential model and provided the best fit of the current COVID-19 data. Hence, these were utilized for predicting the number of COVID-19 cases in India in the future. In view of the active interventions to contain the COVID-19 disease, the polynomial fit and ARIMA models appear to be more useful for predictions of future cases, which shall help the authorities in forecasting the requirement of health infrastructure to contain the effects of the pandemic.