A hybrid of artificial neural network, exponential smoothing, and ARIMA models for COVID-19 time series forecasting

The Auto Regressive Integrated Moving Average (ARIMA) model seems not to easily capture the nonlinear patterns exhibited by 2019 novel coronavirus (COVID-19) in terms of daily confirmed cases. As a result, Artificial Neural Network (ANN) and Error, Trend, Seasonality (ETS) modeling have been successfully applied resolve problems with estimation. Our research suggests that it would be ideal use single ETS or ARIMA for COVID-19 time series forecasting rather than complicated Hybrid combines several models. We compare performance these models using real, worldwide, data period between January 22, 2020 till June 19, 20 2, 2021 which marks two stages, each stage indicating first second wave respectively. discuss various approaches criteria choosing best technique. selected was compared assessment criterion known as Mean Absolute Error (MAE). empirical results show outperform ANN main finding from analysis indicate magnitude increase total cases over is declining percentage change death rate also on decline. shows chosen forecaste are consistent during pandemic. These forecasts encouraging world struggles contain spread COVID-19. This may result social distancing measures mandated governments worldwide.