The Prediction of COVID-19 Pandemic Situation in Indonesia Using SVR and SIR Algorithm
Ainul Yaqin, Majid Rahardi, Ferian Fauzi Abdulloh, Kusnawi Kusnawi, Slamet Budiprayitno, Siti Fatonah · 2022
The COVID-19 epidemic, which initially surfaced at the end of 2019, has since expanded to every corner of the globe and has profoundly impacted all facets of human existence. This case started to emerge in Indonesia at the end of February 2020, and until this point, there has been a spike in new patients. Researchers have run several models and projections for COVID-19 cases in Indonesia, but the results are not yet entirely reliable. Predictions produced at the national level must take into account these variations in patterns because this is likely related to the distinct patterns in each region. In this study, the prediction process will be conducted for cases of COVID-19 by using the SVR algorithm and mathematical models to predict reproduction numbers. SVR analysis to overcome the problem of nonlinearity of data in model formation. The modeling is done based on the SIR model, whose parameters are estimated based on the data. Testing result by using 3 kernels is different on each test, prediction of data cases and the level of mistake room are by using Kernel ’ RBF ‘ with a value of C = 1E3, and gamma = 0.1 with the value of MAPE and MSE respectively are 4.5% and 4.2.