Analysis on the Time Series Forecasting on COVID-19 New Infections Declining in Shanghai, China

Abstract

This research focus on the forecast of the effect of policy influence on COVID-19 in Shanghai, China, during March to May in 2022. By using new infection numbers in 50 days as the train model, four different models were used for forecasting the new infection numbers in future 14 days. Data was published on China Centers for Disease Control and Prevention. Simple exponential smoothing, Holt’s linear method, ARIMA model and ETS model were utilized and compared the accuracy with each other to find the best fit forecast model. As the result, the ARIMA (2,2,2,) model best fits the decline of the new infection. This result needs a condition of similar infectiousness to COVID-19 and the policies and efforts similar to those used to control the outbreak in Shanghai from March to May 2022. This research helps the forecast for policy effects in future epidemiologic disease control.