Prediction of the Spread of Covid-19 in Indonesia Using the SEIRD Model and Hybrid Model with ARIMA Correction

Building an SEIRD segmentation model with life dynamics, estimating its parameters, and utilizing the ARIMA model to determine and predict the difference between the SEIRD model solution and the observed and fitted data constitute the machine learning approach used in this study. We use machine learning techniques to accomplish this. A hybrid method is used to process the newly collected data, using the model predictions and the residuals from the data. The historical values of the infected, recovering, and dying populations divided by the confidence intervals serve as the parameters for the SEIRD model, which in turn serve as parameters for the models' applicability. confidence level range. Long-term and short-term estimates with confidence intervals can be provided by the model, which can assess incoming data in real-time. We tested the model's predictions using actual data on COVID-19 cases in Indonesia. All current models are tested with the maximum allowed deviation during validation using MAE, MSE, MLSE, normalized MAE, and normalized MSE. The findings shown that the features of the current model are highly accurate for infected patients, patients who are recuperating, and patients who have passed away. Governments, business, and policy makers can use the findings of our suggested model to forecast controllable health hazards. This methodology can also be applied to additional research to progress science.