Forecasting the Spread of COVID-19 Using Time Series in Mehriz City, Iran

Coronavirus disease 2019 or COVID-19, which is also called acute respiratory disease NCAV-2019 or commonly called corona, is a respiratory disease caused by acute respiratory syndrome coronavirus-2. Forecasting the number of new cases and deaths today can be a useful step in predicting the costs and facilities needed in the future. This study aims to model and predict new cases and deaths efficiently in the future.

In this article, 9 forecasting techniques were tested on the data of COVID-19 of Mehriz city, Iran as a case study from 2020/02/26 to 2021/12/19 and using the evaluation criteria of mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) of the models were compared.

For cumulative cases of hospitalization, ARIMA, Exponential, Holt-Winters, and STL models performed better and autoregressive neural networks, Theta, and KNN regression showed poor performance. Also, for cumulative mortality cases, KNN regression, Exponential and Theta models have better performance in predicting cumulative mortality cases, and autoregressive neural networks, ARIMA, and cubic spline smoothing showed poor performance.

the best model according to the mentioned evaluation criteria for predicting cumulative cases of hospitalization of COVID-19 is STL model and for cumulative cases of death is the KNN regression model. Also, the autoregressive neural network model has the worst performance among other models, both for hospitalization and death cases. Also, the important point is that the data should be updated in real-time.