Application Of The Hybrid Method Nonlinear Regression With Modified Logistic Growth Model-Exponential Double Smoothing For Forecasting Covid-19 Cases In Indonesia And Armenia

Abstract

Since the first cases of Covid-19 infection were officially recognized and recorded in Indonesia on March 2, 2020 and March 1, 2020 in Armenia, the addition of new cases has not shown any indication of sloping. The relatively high number of new cases indicates that Indonesia has not yet passed the peak of the pandemic. As for Armenia, the addition of new cases indicates a new pandemic peak to be faced. In these conditions, an important question for decision makers (the Government) to find answers to is when and at what level of total cases will the COVID-19 pandemic end in Indonesia or the second wave in Armenia. Forecasting method of Hybrid Nonlinear Regression With Modified Logistic Growth Model - Double Smoothing Exponential and Classical methods is used to predict the Covid-19 cases that occur in Indonesia and Armenia. Based on the model formed, the peak of Covid-19 cases in Indonesia is predicted to occur on November 26, 2020, with the number of cases reaching 5968 cases. As for Armenia, the peak of Covid-19 cases will occur on November 15, 2020, with the number of cases reaching 3098 cases. Covid-19 in both countries is predicted to decline and be constant in 2021. For the country, Indonesia is predicted to begin to stabilize and be controlled in July - August 2021. As for Armenia, Covid-19 is predicted to be under control and approaching 0 cases in February - March 2021. Forecasting models for the Covid-19 cases in Indonesia and Armenia are different, where for the Covid-19 case in Indonesia the Nonlinear Regression with Logistic Growth Model can be used and for the country of Armenia, the Nonlinear Regression with Modified Logistic Growth Model must be used because it has 2 peak cases. Hybrid method is a very good method for optimizing forecast results. Its application in the Covid-19 case in Indonesia and in Armenia shows that the Hybrid method produces a better MAPE value than the Nonlinear Regression with Logistic Growth Model alone or the Exponential Double Smoothing method alone