Performance Analysis of Autoregressive Integrated Moving Average (ARIMA) and ‘earlyR’ Statistical Models for Predicting Epidemic Outbreaks: A Case Study on COVID-19 Data in India

Abstract

The objective of this research article is to analyze the performance of the Autoregressive Integrated Moving Average (ARIMA) model and the ‘earlyR’ statistical model in predicting epidemic outbreaks using COVID-19 data of India. The results of this analysis can help take preventive actions and restrict the spread of the epidemic with the help of projected data. The ‘projections’ module was utilized to generate the epidemic path by aligning the available COVID-19 data from India, distribution of the time interval between successive cases and reproduction number (R0) of the corresponding regions with the ‘earlyR’ epidemic statistical model. The values of (p, d and q) were obtained utilizing the ‘auto.arima’ function, and an ARIMA time series model was created using the ‘forecast’ module to forecast future infected occurrences. The ‘earlyR’ epidemic model yielded a median projected value with an inaccuracy of 35.1 %, while the ARIMA model had a mean error of −1.9 %. A comparison of these methods indicates that the ARIMA model is a superior method compared to the ‘earlyR’ epidemic model in terms of accuracy. HIGHLIGHTS The paper analyzes the performance of the Autoregressive Integrated Moving Average (ARIMA) model and the 'earlyR' statistical model in predicting epidemic outbreaks using COVID-19 data of India Utilized the ‘projections’ module aligning COVID-19 data, time interval between cases, and reproduction number (R0) with the ‘earlyR’ epidemic model. ARIMA model parameters (p, d, and q) were determined using the ‘auto.arima’ function Evaluated real-time data for both models, showing the ARIMA model's effectiveness in predicting COVID-19 cases in India The proposed ARIMA model outperformed existing models, achieving a mean error of −1.9 %, providing valuable insights for health policymakers GRAPHICAL ABSTRACT