Objective: This study aims to forecast the number of deaths and cases in Turkey 150 days after (6 August 2020) the first occurrence of COVID-19 in Turkey. The data used is from 10 March 2020 (the first day has seen of COVID-19 in Turkey) to 15 June 2020 and includes people of all ages from all provinces of Turkey.
Material and Method: The relationship between cases, deaths, patients in intensive care units, intubated patients, and recovered patients, which are observations of COVID-19, was examined with a correlation matrix. Afterward, the ARIMA (0,2,4) model to forecast the number of COVID-19 cases in Turkey and the ARIMA(0,3,1) model to forecast the number of COVID-19 deaths in Turkey were established.
Result: COVID-19 cases were forecasted that there may be 266.692 cases in Turkey on 6 August in the 1st model. Subsequently, a similar forecast has been made on COVID-19 deaths in Turkey on 6 August in the 2nd model. COVID-19 deaths were forecasted that there may be 5718. The p-values of these parameters of models were observed statistically significant (p<0.05). Later, the stationarity of ARIMA models related to these estimates was examined. According to the Augmented Dickey-Fuller (ADF) test results, ARIMA models were stationary and statistically convenient to use (p<0.05). Finally, the Jarque-Bera (JB) test examining the normal distribution assumption was applied and the models were found to be normally distributed.
Conclusions: Consequently, there is an increase in both predicted cases and predicted deaths by the 150th day of COVID-19. These estimates show that the number of cases and deaths will not decrease to zero level until August 6. Factors such as the biological development of the COVID-19 virus, the rate of spread of COVID-19 disease, or the presence of COVID-19 therapy may not cause any increase in these observations. On the contrary, more than expected increase may occur in observed cases.