Prediction of COVID-19 Pandemic Trend by Machine Learning

Abstract

Since 2020, COVID-19 has had a huge impact on people's lives. Including but not limited to economic, educational, medical, and other aspects. During this period, all sectors of society and the government have intervened reasonably, so it is necessary to analyze the data on COVID-19 so far and make scientific predictions. This article starts with the analysis of raw data on COVID-19 from the World Health Organization (WHO). Then four machine learning methods, including the time series model, exponential smoothing model, XGBRegressor method, and polynomial regression model, are leveraged for trend prediction of the original data. The data, with the time ranging from January 2020 to May 2021, is taken as the training object, and then the epidemic in Jul 2021 is used for testing. The number of cases is predicted and compared with real data. It is concluded that the WHO has indeed carried out effective intervention in areas seriously affected by the epidemic and that the time series model predicts the minimum loss value.