Time Series Model for Forecasting the Number of Covid-19 Cases in the
United States of America
Background: Coronavirus disease-19 (Covid-19) had an unprecendented
effect on both nations and health systems. Time series modeling using
Auto-Regressive Integrated Moving Averages (ARIMA) models have been used
to forecast variables extensively in statistics and econometrics.
Objectives: The aim is to predict the total number of Covid-19 cases in
the United States of America using ARIMA models of time-series analysis.
Methods: We used time series analysis to build an ARIMA model of the
total number of cases from January 21, 2020 to August 7, 2020 and used
the model to predict cases in the following 7 days, from August 8, 2020
to August 14, 2020. Hyndman and Khandakar algorithm was used to select
components of ARIMA models. Percentage error was used to evaluate
forecasting accuracy. Results: During the model building period,
4,883,646 cases were diagnosed and during 14 days of validation period
additional 313,502 new cases were added. ARIMA model with (p,d,q)
components of (5,2,1) was used for forecasting. The mean percentage
error of forecast was 0.09% and forecast accuracy was high in the
following week. Conclusion: ARIMA models can ve used to forecast the
total number of cases of Covid-19 patients in the upcoming first week.