COMPARATIVE STUDY OF REPRODUCTION NUMBER ESTIMATION OF COVID-19 IN SIR COMPARTMENTAL MODEL USING KALMAN FILTER AND OPTUNA ALGORITHM FOR DATA-DRIVEN GOVERNANCE

• Main Author: G.P. Situmorang, Adriel
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/78024
• Institution: Institut Teknologi Bandung

The COVID-19 pandemic, which has lasted for the past three years, has significantly disrupted life in society worldwide. Stakeholders in various countries have made considerable efforts to mitigate the pandemic's impact by implementing policies aimed at maintaining stability. One essential approach to observed disease spread in society is through the estimation of the reproduction number ( ) metric. In this thesis, we conducted a comparative study of two methods for estimating R: using the Kalman filter and Optuna. Our research findings reveal that estimation using Optuna on COVID-19 data in Indonesia yields higher accuracy, with a Mean Absolute Percentage Error (MAPE) of 9.5%. This contrasts with the results obtained using the Kalman filter, which has a MAPE of 35%. However, it is important to note that estimation with Optuna requires longer computational time, exceeding 100 seconds, while estimation using the Kalman filter takes less than 10 seconds. The recommendation drawn from this comparison is that Optuna is a more accurate choice for estimating the reproduction number value in COVID-19 data in Indonesia. However, stakeholders need to consider the trade-off between accuracy and computational time. If speed is a critical factor, the use of the Kalman filter remains a good option with lower accuracy but significantly shorter computational time. In the context of data-driven policy-making, the choice of method should be tailored to specific needs and available resources.