7 Discovery of Robust Distributions of COVID-19 Spread

COVID-19 as a pandemic has impacted many lives with continued threat to our way of life. Decision-makers are grappling with the spread without insights into whether they are in a peak or a plateau of the spread. This severely limits resource management in regions severely affected by COVID-19. Analysis of COVID-19 cases and deaths as time-series distributions can yield insights that can aid the efforts in such regions to help estimate the curve of the spread. In this chapter, we use time-series analysis to analyze COVID-19 spread at various locations to study the overall patterns that emerge. We utilize methods including piecewise aggregate approximation (PAA), matrix profiles (MPs), and time-series discretization to identify time periods where the number of cases and deaths reported depicted any major anomalies and where the overall time series seems to be following a trend. Our analysis can prove beneficial to understand the distributions of COVID-19 cases and deaths to quantify the data into different trends to show when the number of cases spiked and when they remained fairly consistent, even when they are in a trend of high number of cases and deaths. This type of trend analysis is particularly useful to compare locations with similar metadata such as number of assisted living communities, population densities, etc., and study if the disease spread is similar or deviating. This can also provide insights into how the policies have had an impact on the spread at these locations.