Fixed link

On Forecasts of the COVID-19 Pandemic

T. J. Mazurek

posted on 11 November 2020

download (621 views, 357 download, 0 comments)

The current COVID-19 pandemic creates seemingly chaotic devastations. Scientists worldwide have failed to predict actual outcomes. A comprehensive study by (Ioannidis, 2020) examines detailed reasons for the failures. A second study (Taleb, 2020) questions the approaches used for pandemic prediction. They find that the pandemic distribution has a fat tail, and proclaim this result has significant implication for pandemic predictions. The latter study used the Generalized Pareto Distribution. This study examines the historical pandemic series using the truncated Pareto I distribution.  It uses the least-squares-fit estimate for this distribution to determine a precise estimate of the tail exponent 1.15. This study uses Monte Carlo techniques to generate synthetic power law series with this exponent. Applying extreme value theory, it compares the characteristics of the pandemic series to that of power law synthetics. The comparisons show excellent agreement between the historical pandemic series and the synthetic ones. This shows that the pandemic’s distribution behaves like the synthetics with this power law, confirming that this distribution is likely to have a fat tail. This study then examines the worldwide countries’ mortality rates in ways similar to those used for the historical pandemic. It finds that this distribution may also have a fat tail with exponent 1.45. This work notes that the historical pandemic distribution is fixed, while it shows that the distribution of the mortality rates continually evolves. The pandemic and mortality distributions’ power law behaviors red-light their high danger, demanding extreme caution for dealing with plagues. However, the fat-tailed distributions cannot make specific predictions. Improved forecasts like those championed by (Ioannidis, 2020) may provide useful guidance for and aid in monitoring the worldwide countries’ individual pandemic responses.