Anticipating the novel coronavirus disease (COVID-19) pandemic

Taranjot Kaur; Sukanta Sarkar; Sourangsu Chowdhury; Sudipta Kumar Sinha; Mohit Kumar Jolly; Partha Sharathi Dutta

We find that the observed trends in the return rate are sensitive to the choice of parameters and can significantly vary between the datasets. High bandwidths reveal the opposite outcome of the return rates for most of the data sets analysed (Figs. 5C-5I ). Since we use Silverman's thumb rule to select the bandwidth, which gives the best fit to the data, therefore, the choice of window size can influence the observed trends (Figs. 5A-5I) . In our work, we find a large size of the rolling window can alter the EWSs analyses and produce misleading estimates for the return rate. For instance, sensitivity analyses show that it is challenging to disentangle accurate signals of an impending transition from the false ones (Figs. 5G and 5P) for a wide range of window sizes and bandwidths. False signals of an alarming situation can deviate from understanding the gravity of any situation and intensity of surveillance needed. Thus, the choice Time-series of triads of the population-weighted total-column NO2 (molecules/cm 2 ) density over the length of the study period for the nine countries considered in this work (depicted by the circular points). The solid curve in each subfigure represents a 10-triad moving window average of the time-series.