Covid-19 forecast using Holt-Winters exponential smoothing

Abstract

Covid-19 has spread throughout the world, including in Indonesia and it is known that the virus is reported to be highly contagious. Indonesia is the fourth-largest population for Covid-19 cases in Asia, while in the world the top three for covid-19 cases are United States, Brazil, and India, so it is inevitable, Indonesia will be greatly felt the spread of the pandemic coronavirus, even estimated to be freed from the coronavirus in a shorter period when compared to other countries with less exposure to COVID-19. In this paper, we study the covid-19 prediction model using Holt-Winters exponential smoothing, for a certain period. This study uses the Covid-19 pandemic data area of Gorontalo, Indonesia, from April 10 to October 13, 2020 (especially total cases). It was found that using Holt-Winters exponential smoothing, the best forecasting model is the one with smoothing parameters α = 0.1 and γ = δ = 0.5 for trend and seasonality respectively, which gives the smallest MAPE value of 6.14.