CoViD19 Meta heuristic optimization based forecast method on time dependent bootstrapped data

Livio Fenga; Carlo Del Castello

In general, predicting the time of a peak conditional to a set of time dependent data is a non trivial task. Often carried out in a multi-tasking fashion, requiring the availability of time and resources, the correct estimation of future turning points can be important in many instances but becomes crucial in the case of epidemic events. These are the typical circumstances when the forecasting exercise is conducted on-line and on a time series exhibiting a small sample size. However, under these conditions, the problem might become particularly complicated since statistical methods usually employed for this purposes -for example of the type hidden Markov (see, e.g. Hamilton (1989) and Koskinen andÖller (2004) ) or non parametric (see, e.g., Delgado and Hidalgo (2000) ) models -not only are very demanding in terms of building and tuning procedures, but typically require the availability of a "long" stretch of data. In addition to that, the time series related to epidemics usually show highly non-linear dynamics, which, if not pre-processed, make them not suitable for standard linear models. On the other hand, attempting to fit non-linear models -e.g. of the type Self Exciting Threshold Autoregressive (see, for example Clements et al. (2003) ) or artificial neural network (Hassoun et al. (1995) ) -it is not a viable options, due to the above mentioned sample size issues. In any case, when an ill-tuned model is fitted on a time series, reliable outcomes should not be reasonably expected. Therefore, an approach able to perform under the above outlined conditions, is proposed. In essence, the problem is solved by building a unified framework in which two powerful techniques -belonging to two different branches of computational statistics -are sequentially employed to lower the amount of uncertainty embedded in the observed data and to find a (possibly global) Page 1 of 10 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.