Using of Box and Jenkins and Artificial Neural Networks for Modeling The Covid-19 Epidemic in Iraq

ABSTRACT

We aim by this research to use mathematical methods to model the Coved-19 epidemic in Iraq by comparing time series by using box & Jenkins model and artificial neural networks. The infections, cures and deaths data were used for the period from 24/2/2020 to 30/11/2020. The study found a tendency in the numbers of infections and cures using the Box & Jenkins model to rise, while the numbers of deaths tended to stabilize. Artificial neural networks, using the MLP algorithm, have found a tendency to number of infections by decline and cures to rise, while deaths numbers tended to decrease and then to stability. In addition, the study found that the forecasting of the numbers of infections was more accurate using artificial networks, while the forecasting of the numbers of cures was more accurate in the Box Jenkins model and the forecasting of death numbers was at the same level of accuracy in the trade-off between the two methods. The study recommends to sue the artificial networks to forecast the number of infections and deaths and the use of the Box Jenkins model to forecast cures. In addition, the study recommends the use of these mathematical methods to help decision makers respond to the epidemic. And also recommends to conduct another study using other techniques for artificial networks as an algorithm extreme learning machines (ELM). The study also recommends a survey of habits associated with the spread of the epidemic, such as social distancing and other, linking them with the numbers of infections, cures and deaths to reach a protocol specific to Iraq based on accurate mathematical and scientific foundations. © 2022 American Institute of Physics Inc.. All rights reserved.