Real-time forecasts and risk assessment of novel coronavirus (COVID-19) cases: A data-driven analysis

Tanujit Chakraborty; Indrajit Ghosh

We made some simplifying assumptions to carry out the analysis of COVID-19 datasets. The assumptions are listed as follows: (a) the virus mutation rate are comparable for different countries; (b) the recovered persons will achieve permanent immunity against COVID-19; (c) we ignore the effect of climate change (also spatial data structures) during the shortterm predictions. Along in this line, we presented two different approaches to deal with two inter-connected problems on COVID-19. In the first problem of short-term predictions for COVID-19 outbreak in five countries, we proposed a hybrid methodology combining ARIMA and WBF models. In the second problem of risk assessment, we found some important factors affecting case fatality rates of COVID for 50 highly affected nations. However, there may exist a few more controllable factor(s), and some disease-based characteristics that can also have an impact on the value of CFR for different countries, can be regarded as future scope of the study.