Application of Machine Learning Techniques in Building a Time Series Forecasting Model of the Number of COVID-19 Infections

Abstract

Building predictive models is a common scientific task that enables humans to plan or detect anomalies. In particular, much scientific research related to Covid-19 has been carried out. Many aspects of this pandemic are included in the development of infection prediction models. In this project, we focus on learning about time series prediction methods, combined with various data processing techniques and machine learning algorithms to solve the problem of predicting the number of cases infected with Covid19. Through the experimental process, we initially propose the Extreme Learning Machines model with respective MSE and sMAPE of 33.10(9) and 13%. Throung thi study, many new features of the data and models were discovered and allowed us to discuss the potential future development avenues of this problem in more depth.