Assessing the Global Tendency of COVID-19 Outbreak

Qinghe Liu; Zhicheng Liu; Junkai Zhu; Yuhao Zhu; Deqiang Li; Zefei Gao; Liuling Zhou; Junyan Yang; Qiao Wang

The copyright holder for this preprint . https://doi.org/10.1101/2020.03. 18.20038224 doi: medRxiv preprint To further analyze the characteristics of epidemic transmission among different countries or regions, we took the five parameters, i.e. , , 1 , 2 , , of the SEIR model obtained in the simulation as feature vectors. Then the BIRCH (Balanced Iterative Reduction and Clustering Using Hierarchies) algorithm was used to perform clustering based on these feature vectors to retrieve countries or regions with similar epidemic patterns, see Figure 7 and Appendix 3. The average of parameter in each cluster is shown in Table 4 . To visualize the proximity relationship of epidemic characteristics of these countries or regions, t-SNE (t-distributed stochastic neighbor embedding) algorithm is used to reduce the dimension of data to a 2-dimensional plannar, while preserving sample proximity in the 5-dimentional parameter space, see Figure 7 .