A team at King’s Business School, working with a colleague at the University of Strathclyde have produced a draft paper for the Qatar Centre of Global Banking and Finance designed to assist central banks by applying more complex statistical models to case data on the pandemic in six of the world’s most affected countries. As well as providing useful information on the global picture, the methodology can also be applied to countries’ own available data.
The team will publish updated forecasting on a regular basis.
In this paper we provide methods for forecasting the evolution of the Covid-19 pandemic across the globe. In particular, we use statistical, time-series, approaches, in order to forecast the rate of growth of the confirmed cases of Covid-19 pandemic in the most affected countries.
The paper provides both a set of forecasts which we plan to regularly update, as well as a set of realistic methods that can be used to forecast the pandemic in any country or region, without many strict assumptions.
We use non-linear curve fitting models, including logistic and spline functions, and also more elaborate parametric neural network models. We find more complex models to have better forecasting performance.
Download the paper [PDF].