Comparison of auto ARIMA and auto SARIMA performance in COVID-19 prediction

Abstract

Forecasting the number of COVID-19 cases is crucial in effectively managing and reducing the consequences of the current global health crisis. The objective of this study is to observe the performance comparison between Automatic Autoregressive Integrated Moving Average (Auto-ARIMA) and Automatic Seasonal Autoregressive Integrated Moving Average (Auto-SARIMA). COVID-19 data was retrieved from Malaysia's Ministry of Health open data source from 24th February 2021 to 2nd January 2022. Results revealed that the ARIMA (3,1,2) model demonstrated superior predictive precision in contrast to the SARIMA (0,1,1) (1,0,1,7) model. The ARIMA model yielded an accuracy of 87.81 percent. On the other hand, the SARIMA model produced an accuracy of 78.80 percent. The study's results indicate that utilizing the ARIMA model could potentially enhance the accuracy of COVID-19 case predictions and support decision-making in managing the COVID-19 pandemic in Malaysia.