Developing forecasting model for future pandemic applications based on COVID-19 data 2020-2022

Area of Science:

• Epidemiology
• Data Science
• Public Health

Background:

• Accurate COVID-19 forecasting is vital for effective pandemic control.
• Existing models struggle with the complex linear and non-linear patterns in COVID-19 data, leading to inaccuracies.
• There is a need for more precise and efficient prediction methods.

Purpose of the Study:

• To propose a hybrid ARIMA-SVM model for enhanced COVID-19 forecasting.
• To evaluate the performance and improvements of the hybrid model against standalone ARIMA and SVM models.
• To validate the proposed model's accuracy and efficiency using statistical metrics.

Main Methods:

• Development of a hybrid ARIMA-SVM model.
• Comparative analysis using statistical measurements: Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE).
• Empirical testing on three real-world COVID-19 datasets from Malaysia.

Main Results:

• The hybrid ARIMA-SVM model consistently produced lower MSE, RMSE, MAE, and MAPE values compared to ARIMA and SVM models.
• The proposed model demonstrated superior accuracy and efficiency in predicting COVID-19 cases on both training and testing datasets.
• Significant error reduction percentages were achieved, with maximum improvements of 73.12% (MAE), 74.6% (MAPE), 90.38% (MSE), and 68.99% (RMSE).

Conclusions:

• The hybrid ARIMA-SVM model offers a more accurate and efficient approach to COVID-19 forecasting.
• This model provides a valuable tool for public health authorities to improve pandemic monitoring and prevention strategies.
• The enhanced prediction performance validates the hybrid model as an effective method for managing infectious disease outbreaks.