Forecasting Covid-19 Infections in Ho Chi Minh City Using Recurrent Neural Networks

ABSTRACT

Coronavirus disease (Covid-19) has caused negative impacts on the economy, society and lives of people in Vietnam, especially in Ho Chi Minh City. Forecasting daily new Covid-19 infections is essential and important for prevention and social distancing purposes. Recurrent neural networks have been intensively used to process sequential data like voice, text, video and time series recently. In this paper, we present the forecasting models to predict new Covid-19 infected cases in Ho Chi Minh City using different recurrent neural networks (RNN). The experimental results show that the bidirectional long short-term memory network obtains better performance than the other models based on three statistical assessment criteria, the mean absolute error (MAE), symmetric mean absolute percentage error (sMAPE) and root mean square error (RMSE). The forecasting performance is also verified on the different forecasting horizons and multiple test runs. © 2021, Springer Nature Singapore Pte Ltd.