In collaboration with George Atsalakis from the Technical University of Crete, Data Analysis and Forecasting Lab, we modelled the development of the coronavirus in France to predict its path. You can see in blue the actual cases and in purple the predicted path. The model fits reality quite well with an accuracy of 80%. The volatility of the spikes is to be expected and to some small extent could be linked to daily numbers not reflecting all cases and added to another day.
The algorithm is using the daily official cases that are collected by the John Hopkins University and then confirmed via Wikipedia. The algorithm anticipates roughly 10 times more real cases to the ones announced and can be updated every three days for adjustments. Contrary to epidemiological studies that expect to see a flattening of the sigmoid curve, we focus on the peak of the bell shaped curve and based on the fit of the data and the symmetry of this curve we can predict that the end of the pandemic using the time it took to get to the peak in the first place.
The purpose of this effort is to give to states and the business world a timeline to the end of the pandemic so they can plan the next day. This model was used in the case of Greece with a remarkable success as it predicted as early as 28 March the peak that occurred on the 2nd of April against expectations of leading world experts that the peak would be on the 10th of April (there the accuracy was 90% as the data were smoother).
For France as we can clearly see, the peak of the virus is behind us and there is a clear downwards trend expectation by the model that is also confirmed by the number of actual new cases, opening the way of the next day (provided this trend continues, i.e. the public respects the government’s restrictions until 11th of May). The model shows that by the end of the month the new cases will be very few if people continue observing the lock down rules and at the 11th of May when the French president announced that the economy will reopen gradually the cases will be even less and manageable by the public health system and should remain so, provided again, that the public respects the renewed social distancing rules and protection measures that will be announced by the state.