The Prediction Of COVID-19 Pandemic Using The Fast Fourier Transform And Gaussian Mixture Model

Since the first novel coronavirus infection case was found in 2019,due to the characteristics of novel coronavirus with high infectivity and concealment,it has spread to all countries in the world successively,resulting in a worldwide pandemic.The sudden outbreak of COVID-19 has caused a major impact on the global medical system and a huge negative impact on economic operation.The prevention and control of the epidemic is facing huge challenges.In this paper,first of all,the world health organization statistics of the number of newly increased patients every week around the world and the death toll to integrate for panel data,descriptive statistics of the panel data analysis,found that according to the geographical position could be divided into six regional to the global development trend of the new champions league the irrationality of clustering,it did not take into account the time sequence of the correlation.Secondly,the boxplot is used to measure the autocorrelation within the time series,and the changes of the autocorrelation coefficient in six regions in the time dimension are discussed.By analyzing the autocorrelation coefficient and cross correlation coefficient of the time series,the feasibility of Fourier transform to eliminate the autocorrelation of the time series is proved.Thirdly,Fourier transform was used to eliminate the autocorrelation of time series,and Gaussian mixture model was used to cluster according to BIC criterion.The characteristics of each class were analyzed through the clustering results,including the parameters of Gaussian mixture model and the correlation coefficient between classes.Finally,K-means and growth mixture model are compared with Gaussian mixture model,and the applicability and advantages of each model are analyzed and compared.In this paper,the development trend of COVID-19 epidemic was predicted by Gaussian mixture model.Under the assumption that under the condition of certain constraints was established,by building a consistent with actual situation and reasoning logic of statistical model,the short-term prediction of epidemic trend has certain accuracy,prediction results can also evaluate the effectiveness of prevention and control measures as an important reference,and benefit to the medical and health resources fully and reasonable allocation,Avoid the strain and waste of medical resources.Using mathematical and statistical tools to study the spread,control and development trends of sudden epidemics is the support and promotion for the continuous progress of public health discipline in China.