ANALYSIS OF THE SPREAD OF COVID-19 IN INDONESIA USING SIR AND SIR-F MODELLING WITH MACHINE LEARNING CONCEPT IMPLEMENTATION

• Main Author: Nararia Rahman, Fadel
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55606
• Institution: Institut Teknologi Bandung

The COVID-19 virus pandemic in Indonesia has been going on since March 2020 and is still ongoing with conditions that need to be watched out for. This can be seen from the distribution of additional daily active cases in Indonesia which is still changing dynamically. Alternative solutions that can help to analyze the prevention of the spread of the virus are modelling and simulating the spread of cases to estimate a picture of the pandemic conditions that might occur in Indonesia. A common and widely used epidemiology-based model is SIR modelling, which classifies individuals affected by a pandemic into a number of compartments. Using this model and utilizing the concept of machine learning technology, the modelling process can be carried out more efficiently and accurately. In this final project, two models were developed, namely SIR and one of its derivatives, SIR-F, based on the concept of machine learning to estimate and simulate various virus spread scenarios. There are 3 scenarios developed to be analyzed, namely a scenario without a vaccination program, a vaccination program with health protocols that are adhered to, and a vaccination program that is not followed by a health protocol. Based on the simulation scenario, it was found that the vaccination program could actually have a positive impact on efforts to deal with the COVID-19 pandemic more effectively when compared to a scenario without vaccination. Meanwhile, if the vaccination program is not supported by adequate health protocols, then vaccination will not have any impact on the prevention efforts. These results apply uniformly to the results of both the SIR and SIR-F models. Overall, it can be concluded that the developed model can carry out all its functions as needed, with the level of accuracy through the MAPE metric reaching 0.412 for the SIR model and 0.022 for the SIR-F model.