Performance of Different Models of Machine Learning in Predicting the COVID-19 Pandemic

摘要

COVID-19 outbreaks experienced explosive growth worldwide in late 2019 and early 2020. In order to control the spread of the epidemic, many researchers used different machine learning models to predict the trend of the epidemic. This paper tested the performance of some common machine learning models to predict the epidemic, so as to provide a basis for future researchers to choose models. This paper used different models of machine learning, including the Susceptible Infected Recovered Model, classic machine learning of the Linear Regression, Polynomial Regression, k-Nearest Neighbor, Logistic Growth Model and the Long Short-Term Memory model of deep learning. Based on the data of the United States, Japan, Wuhan, China, we intended to predict the COVID-19 trend in these countries with the parameters of the machine learning and intuitive chart to measure the prediction results of different model; also, we put into consideration the degree of each country to the attention of the outbreak as well as control mode and analyze the different models to predict the tendency of the epidemic development rationality and validity.