COVID-19 Death and Case Numbers Forecasting with ARIMA and LSTM Models

Abstract

The Covid-19, which quickly turned into a pandemic, has not yet been fully controlled despite the vaccines developed. The nearly two-year period of struggling with the pandemic has caused a global economic crisis. Many countries have lifted the restrictions they have applied in the fight against the pandemic to get rid of this crisis. Despite the vaccines, the pandemic still poses a great danger, and it remains unclear when both the pre-pandemic life can be returned, and the economic crisis can be brought under control. For this reason, the correct analysis of the picture that emerged in line with the policies followed so far is still an essential problem in accurately predicting the future course of the pandemic. In this study, Covid-19 estimation is made with Auto Regressive Integrated Moving Average and Long-Short-Term Memory models using daily case and death numbers for Germany, France, Italy, Ireland, Poland, Russia, and Turkey. Root mean square error, mean absolute percentage error, mean absolute error, Adjusted R2, Akaike Information Criterion, and Schwarz Information Criterion metrics are used in model selection. The results showed that Auto Regressive Integrated Moving Average and Long-Short-Term Memory models could be used to predict the number of COVID-19 deaths and cases. Furthermore, it has been seen that the prediction success of the Long-Short-Term Memory models for the countries considered is higher than the Auto Regressive Integrated Moving Average models.

Keywords

• ARIMA
• Covid-19
• Forecasting
• LSTM

References

1. [1].Khan F, Gupta R. 2020. ARIMA and NAR based prediction model for time series analysis of COVİD19 cases in IndiaJournal of Safety Science & Resilience, 12-18.
2. [2].Özen N, Saraç S. ve Koyuncu M. 2021 Prediction of COVID-19 Cases in the United States of America with Machine Learning Algorithms, European Journal of Science and Technology Special Issue 22, pp. 134-139.
3. [3].Sevli, O. & Başer, V. G. 2020 Machine Learning Based Case Estimation Using Prophet Model with Time Series Data for Covid-19 OutbreakEuropean Journal of Science and Technology No. 19, 827-835.
4. [4]. Awan T. M., Aslam F. 2020. Prediction of daily COVID-19 cases in European counties using automatic ARIMA model J Public Health Res. 9(3),1765.
5. [5].Kırbaş İ., Sözen A., Tuncer A.D., Kazancıoğlu F.Ş. 2020 Comparative analysis and forecasting of COVID-19 cases in various European countries with ARIMA, NARNN and LSTM approaches Chaos Solitons Fractals 110015
6. [6].Hernandez-Matamoros, A., Fujita, H., Hayashi, T., & Perez-Meana, H. 2020. Forecasting of COVID19 per region using ARIMA models and polynomial functions. Applied Soft Computing, 96, 106610.
7. [7].Roy, S., Bhunia, G. S., & Shit, P. K. 2021. Spatial prediction of COVID-19 epidemic using ARIMA techniques in India. Modeling earth systems and environment, 7(2), 1385-1391.
8. [8]. Moftakhar, L., Mozhgan, S. E. I. F., & Safe, M. S. 2020. Exponentially increasing trend of infected patients with COVID-19 in Iran: a comparison of neural network and ARIMA forecasting models. Iranian Journal of Public Health.

9. [9].Alzahrani, S. I., Aljamaan, I. A., & Al-Fakih, E. A. 2020. Forecasting the spread of the COVID-19 pandemic in Saudi Arabia using ARIMA prediction model under current public health interventions. Journal of infection and public health, 13(7), 914-919.
10. [10].Elsheikh, A. H., Saba, A. I., Abd Elaziz, M., Lu, S., Shanmugan, S., Muthuramalingam, T., ... & Shehabeldeen, T. A. 2021. Deep learning-based forecasting model for COVID-19 outbreak in Saudi Arabia. Process Safety and Environmental Protection, 149, 223-233.
11. [11]. Ala’raj, M., Majdalawieh, M., & Nizamuddin, N. 2021. Modeling and forecasting of COVID-19 using a hybrid dynamic model based on SEIRD with ARIMA corrections. Infectious Disease Modelling, 6, 98-111.
12. [12].Eroğlu Y. 2020. Forecasting Models For Covid-19 Cases of Turkey Using Artificial Neural Networks and Deep Learning,Journal of Industrial Engineering 31(3), 354-372.
13. [13]. URL https://github.com/owid/covid-19-data/tree/master (accessed at 12.11.2021).
14. [14]. Box, G.E.P. and Jenkins, G.M. 1970. Time Series Analysis: Forecasting and Control. San Francisco: Holden-Day.
15. [15]. Özmen A., 1986 Zaman Serisi Analizinde Box-Jenkins Yöntemi ve Banka Mevduat Tahmininde Uygulama Denemesi, Anadolu Üniversitesi Yayınları, 207, Eskişehir. (Özmen A.Box-Jenkins Method in Time Series Analysis and Application Trial in Bank Deposit Estimation, Anadolu University Press, 207, Eskisehir 1986)
16. [16 ]. Akdi, Y. Zaman Serileri Analiz, Genelleştirilmiş2. Baskı, Gazi Kitabevi, Ankara.2010 (Akdi, Y. Time Series Analysis, Generalized 2nd Editionı, Gazi Bookstore, Ankara.2010)
17. [17].Hamzaçebi, C. ve Kutay, F., 2004.Electric Consumption Forecasting of Turkey Using Artificial Neural Networks up to Year 2010”, Journal of The Faculty of Engineering and Architecture of Gazi University, (19), No 3, 227-233.
18. [18]. Hochreiter & Schmidhuber, 1997. Long Short-Term Memory, Neural Computation 9(8):1735-1780.
19. [19].Kara A. 2019. Global Solar Irradiance Time Series Prediction Using Long Short-Term Memory NetworkJournal of Science, Gazi UniversityGU J Sci, Part C, 7(4): 882-892.
20. [20]. T. W. C. B. Aya Abdelsalam Ismail, 2018. Improving Long-Horizon Forecasts with Expectation-Biased,arXiv:1804.06776 [cs.LG]
21. [21]. M. Yuan, Y. Wu, L. Lin, 2016.Fault Diagnosis and Remaining Useful Life Estimation of Aero Engine Using LSTM Neural Network, IEEE International Conference on Aircraft Utility Systems (AUS), 135–140.
22. [22]. Olah, C. 2015. Understanding LSTM Networks. August 7,2021, colah.github.io: colah.github.io/posts/2015-08-Understanding-LSTMs
23. [23]. Emang, D., Shitan, M., Abd Ghani, A. N., & Noor, K. M. 2010. Forecasting with univariate time series models: A case of export demand for peninsular Malaysia's moulding and chipboard. Journal of Sustainable Development, 3(3), 157.