FORECASTING COVID-19 IN INDONESIA WITH VARIOUS TIME SERIES MODELS

In this study, Covid-19 modeling in Indonesia is carried out using a time series model. The time series model used is the time series model for discrete data. These models consist of Feedforward Neural Network (FFNN), Error, Trend, and Seasonal (ETS), Singular Spectrum Analysis (SSA), Fuzzy Time Ser...

Full description

Saved in:
Bibliographic Details
Published inMedia statistika (Online) Vol. 15; no. 1; pp. 83 - 93
Main Authors Darmawan, Gumgum, Rosadi, Dedi, Ruchjana, Budi Nurani, Pontoh, Resa Septiani, Asrirawan, Asrirawan, Setialaksana, Wirawan
Format Journal Article
LanguageEnglish
Published Universitas Diponegoro 27.07.2022
Subjects
covid-19
ffnn
fts
garma
singular spectrum analysis
Online AccessGet full text

Cover

More Information
Summary:In this study, Covid-19 modeling in Indonesia is carried out using a time series model. The time series model used is the time series model for discrete data. These models consist of Feedforward Neural Network (FFNN), Error, Trend, and Seasonal (ETS), Singular Spectrum Analysis (SSA), Fuzzy Time Series (FTS), Generalized Autoregression Moving Average (GARMA), and Bayesian Time Series. Based on the results of forecast accuracy calculation using MAPE (Mean Absolute Percentage Error) as model evaluation for confirmed data, the most accurate case models is the bayesian model of 0.04%, while all recovered cases yield MAPE 0.05%, except for FTS = 0.06%. For data for death cases SSA and Bayesian Models, the best with MAPE is 0.07%.
ISSN:1979-3693
2477-0647
DOI:10.14710/medstat.15.1.83-93