SARIMA MODELLING OF DAILY LABORATORY CONFIRMED CASES OF CORONAVIRUS IN NIGERIA Ette Harrison Etuk Department of Mathematics Rivers State University, Port Harcourt, Nigeria Email: ettetuk@yahoo.com, ettehetuk@gmail.com, etuk.ette@ust.edu.ng
ABSTRACT
This study is an attempt to model daily confirmed cases of coronavirus in Nigeria. A time plot of the series shows an upward trend with some seasonality. It is tested for unit test and is shown to be non-stationary. Its difference shows evidence of stationarity. The correlogram of the difference shows significant spikes at the partial autocorrelation function at lags 1 and 12 and at its autocorrelation function at lags 1 and 13, with the lag 13 spike surrounded by spikes of comparable lengths in the same direction. This suggests an autoregressive fit of lags 1 and 12 and a moving average fit of lags 1, 13 and 14. A fit of the model shows that only the moving average lags are significant. A more specific SARIMA(0, 1, 1)x(0, 0, 1)13 model is fitted to the series. This shows that the series may be regarded as a SARIMA(0, 1, 1)x(0, 0, 1)13 case.
Key Words: Covid-19 pandemic, SARIMA modelling
