Forecasting the COVID-19 Pandemic in Bangladesh Using ARIMA Model

The effects of the coronavirus disease in 2019 are visible in every corner of the globe. The public health system is mostly affected, and the economic and social crises are also increasing day by day. Due to the widespread nature and the unavailability of drugs or vaccines for this pandemic, it is urgent to predict the COVID-19 infected cases to handle the situation more efficiently. Time series prediction is a crucial technique of the machine learning domain to deal with the issue. This research aims to predict the number of daily confirmed COVID-19 cases for a successful time. To forecast COVID-19 instances in Bangladesh, we use the Autoregressive Integrated Moving Average (ARIMA) model. The experimental results show that the estimated best models are: ARIMA(3,1,0) with drift, ARIMA(3,1,2) with drift, ARIMA(5,1,0) perform significant predictions on three different kinds of COVID-19 datasets.