Covid-19 Modelling in South Korea using A Time Series Approach

Abstract

COVID-19 pandemic has confirmed to spread fast for more than two hundred countries in the world. Therefore, to give accurate information to their government, many researchers try to forecast how many people are infected. South Korea is one of the countries affected by the virus which the first case of COVID-19 was announced on January 20th, 2020. Furthermore, this research tries to model the incidence of COVID-19 cases in South Korea which until now continues to increase by using a nonparametric time series from Neural Networks such as MLP, NNETAR, GRNN, and ELM. By evaluating these simple models, it is hoped that suitable data patterns can be found to describe the development of this pandemic in South Korea. The more important thing is the result can later be used as basic public information about future conditions. Many countries classified the COVID-19 sufferer as confirmed, recovered and death cases. The overall result shows that MLP with two hidden layeres performs best to forecast the confirmed, recovered, and death cases cause this model produces the lowest RMSE and MAE values.