【摘要】：目的探討時間序列分析中的自回歸求積移動平均模型(ARIMA)在南通市流感樣病例(ILI)發(fā)病趨勢預(yù)測中的應(yīng)用。方法收集"中國疾病預(yù)防控制系統(tǒng)"中2009年7月-2013年11月共238周流感樣病例的發(fā)病監(jiān)測數(shù)據(jù),建立時間序列數(shù)據(jù)庫,對每周流感樣病例的發(fā)病人數(shù)進行ARIMA模型擬合,利用模型對2009年7月-2013年11月的周數(shù)據(jù)進行外部預(yù)測,并對2014年2月-2014年3月各周的流感樣病例的發(fā)病情況進行前瞻性預(yù)測。結(jié)果構(gòu)建流感樣病例周發(fā)病數(shù)的ARIMA(1,1,1)模型為(1-0.806B2)(1-B)ln Xt=(1-0.958B2)∝t,其中B代表后移算子,Xt代表ILI周發(fā)病數(shù),εt為隨機誤差。外部預(yù)測的預(yù)測值和實際值基本相符,相對誤差較小。前瞻性預(yù)測結(jié)果符合流感樣病例的流行特征。結(jié)論該模型能較好的模擬并預(yù)測南通市流感樣病例的發(fā)病趨勢。
[Abstract]:Objective to explore the application of autoregressive quadrature moving average model (Arima) in predicting the trend of influenza like case (ILI) in Nantong city. Methods from July 2009 to November 2013, the surveillance data of 238 weeks of influenza like cases in China Disease Prevention and Control system were collected, and a time series database was established. The Arima model was used to fit the number of influenza-like cases per week. The model was used to predict the weekly data from July 2009 to November 2013, and the incidence of influenza-like cases from February 2014 to March 2014 was forecasted. Results the Arima model was (1-0.806B2) (1-B) ln Xt = (1-0.958B2) 鈭，

本文編號：2127026