%0 Journal Article
%T 我国COVID-19疫情时空演变特征与趋势研究——基于Spatial Markov Chain和STL时间序列模型<br>Study on Trend and Characteristics of Spatio-Temporal Evolution of COVID-19 Epidemic in China—Based on Spatial Markov Chain and STL Time Series Model
%A 张小英
%A 巫细波
%J Hans Journal of Data Mining
%P 8-19
%@ 2163-1468
%D 2022
%I Hans Publishing
%R 10.12677/HJDM.2022.121002
%X 基于2020年1月24日~2021年3月4日省级行政区的COVID-19肺炎现有确诊病例数、累计确诊病例数、治愈数等统计数据，本文采用GIS分析、空间马尔科夫链、STL时间序列模型等方法从时间和空间两个维度分析我国新型冠状病毒肺炎（简称COVID-19）疫情时空演变特征。研究发现：① 全国COVID-19疫情大致分为“大规模快速爆发期、全国严格防控期、全国抑制期、局部复发期、常态化防控期、二次局部复发期、常态化防控期”等7个阶段，绝大部分省份疫情变化特征与全国总体情况相一致，个别省份因局部地区爆发疫情导致疫情变化特征表现出一定差异性。② 湖北春运期间的大规模人口流动是导致疫情在全国范围内快速扩张的主要原因，湖北封城之前14天的百度迁徙规模指数与全国各省市区的累计确诊数(到2月11日)呈现显著正相关。③ 采用马尔科夫链模型分析各省市COVID-19疫情现存确诊数的结果显示，4种疫情风险(高感染、较高感染、较低感染、低感染)保持不变的概率较为稳定且空间分布相对固定，平均向下转移的概率高于向上转移概率，反映了我国COVID-19现存确诊数量下降的趋势。④ 采用STL时间序列趋势分析法分析我国及典型省市的每周现存确诊病例数变化特征及趋势，我国中后期现存确诊数总体呈现平稳趋势，新增确诊病例主要来自境外输入病例，输入性防控将是我国疫情防控重点。<br />
Based on statistical data including the number of active cases, accumulated confirmed cases and cured case of Novel Coronavirus Pneumonia (COVID-2019) in the provincial administrative region from January 24, 2020 to March 4, 2021, this paper uses methods including Geographic Information System (GIS) analysis, Spatial Markov Chain, STL time series model and so on to analyze spatial and temporal evolution characteristics of Novel Coronavirus Pneumonia in China. The study found that: ① The COVID-19 epidemic in China can be roughly divided into seven stages including large-scale rapid outbreak period, national strict prevention and control period, national suppression period, local recurrence period, normalized epidemic prevention period, second local recurrence period and normalized epidemic prevention period. The epidemic change characteristics of most provinces are similar to the nationwide’s situation. Due to the outbreak of the local epidemic, the characteristics of epidemic changes in individual provinces show some differences. ② The high population mobility from Hubei during the Spring Festival transportation period are the main reasons for the rapid expansion of the epidemic. The Baidu Migration Scale Index for the 14 days prior to Hubei closure was significantly correlated with the cumulative diagnosis of provinces (by February 11). ③ The analysis results of the Markov chain model for active cases of COVID-19 in provinces show that the four epidemic risks (high infection, higher infection, lower infection, and low infection) have a relatively stable probability and a relatively fixed spatial distribution, and the average probability of downward transfer is significantly higher than the probability of upward transfer, reflecting the downward trend in the number of active cases of COVID-19 in China. ④ Based on the STL time series trend analysis method, this paper analyzes the change characteristics and trend of weekly con-firmed cases in China and typical provinces. The number of
%K COVID-19，时空演变，空间马尔科夫链模型，STL时间序列模型<br>COVID-19
%K Spatio-Temporal Evolution
%K Spatial Markov Chain Model
%K STL Time Series Model
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=47402