Forecasting Short-Term Covid-19 Cumulative Cases with Prophet Model: An Example from G7 Countries
Keywords:
Forecasting, Prophet Model, Covid-19, Machine LearningAbstract
The Covid-19 pandemic, which emerged in Wuhan, China, in December 2019, has become unpreventable worldwide after spreading rapidly across the country. The unpreparedness of coun-tries for the pandemic has negatively affected many areas, especially the health sector, and has caused a struggle with a lack of capacity, resources, and commitment in general. Therefore, predic-tions have become essential to understand the mechanism of infectious disease spread. In literature, while there are many models for obtaining infectious disease predictions, machine learning models are also widely used today. The Prophet model, as one of these models, consists of three main com-ponents: a trend (logistic) function that captures non-periodic changes, a seasonality (Fourier) func-tion that captures periodic changes, and a holiday function that captures irregular periods. The study aims to obtain five-day short-term predictions and prediction intervals using the Prophet model with the cumulative data of G7 countries. Additionally, the RMSE statistic is used to compare the performance of the predictions. As a result, relatively better results are obtained for Canada and Germany with low RMSE values. In conclusion, the Prophet model provides researchers and poli-cymakers with high-performance results with low RMSE values in short-term predictions.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2023 Journal of Economics and Business Issues
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
All articles published in this journal are licensed under a
Creative Commons Attribution-NonCommercial 4.0 International License