PENERAPAN MODEL LEAST SQUARE SUPPORT VECTOR MACHINE (LSSVM) UNTUK PERAMALAN KASUS COVID-19 DI INDONESIA

Lutfi Ardining Tyas (Jurusan Matematika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Universitas Jember, Jl. Kalimantan No. 37 Kampus Tegalboto, Jember, Indonesia 68121)
I Made Tirta (Jurusan Matematika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Universitas Jember)
Yuliani Setia Dewi (Jurusan Matematika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Universitas Jember)

Jurnal Gaussian
Vol 12, No 2 (2023): Jurnal Gaussian

Forecasting is about predicting the future based on historical data and any information that might affects the forecasts. This article applies the LSSVM model to forecast Covid-19 cases in Indonesia. The purpose of this study is to find out how the LSSVM model applied and the model performances for forecasting Covid-19 cases in Indonesia, using time series data and the factors that influence it, as input features. The factor data used in this study are mobility data and daily fully vaccinated data. The research has three main objectives; first, calculate the correlation between confirmed cases data and past data (lag) of mobility and vaccination. Second, is the selection of input features based on the highest correlation coefficient value of each variable. Third, do LSSVM modeling and Covid-19 case forecasting with the optimal model. RBF kernel and grid-search algorithm with 10-fold cross-validation are used to tune model parameters. The results show that the LSSVM model provides good performance for Covid-19 forecasting and the optimal LSSVM model for forecasting Covid-19 cases in Indonesia is using time lag 14 for the mobility factor and time lag 24 for the vaccination factor.