论文部分内容阅读
目的探讨应用自回归滑动平均混合模型(autoregressive integrated moving average,ARIMA)预测婴儿死亡率的可行性。方法运用SPSS 16.0对1991-2012年山西省妇幼卫生年报婴儿死亡率建立ARIMA模型,用所建模型比较预测值与实际值差异,并预测2013-2015年山西省婴儿死亡率。结果模型ARIMA(1,2,0)较好地拟合了既往时间段的婴儿死亡率的时间序列,模型自回归参数AR1=-0.754,P<0.01,有统计学意义,赤池信息准则(AIC)=68.213,许瓦兹贝叶斯准则(SBC)=70.204,模型残差为白噪声(P>0.05),模型数学函数式为^Yt=0.067+1.246Yt-1+0.508Yt-2-0.754Yt-3,利用模型预测2013-2015年婴儿死亡率分别为4.77‰、4.32‰、3.96‰。结论 ARIMA模型能够较好地拟合婴儿死亡率的时间变化趋势,并用于短期预测未来婴儿死亡率。
Objective To explore the feasibility of using autoregressive integrated moving average (ARIMA) to predict infant mortality. Methods SPSS 16.0 was used to establish the ARIMA model for infant mortality rate of MCH in Shanxi Province from 1991 to 2012. The model was used to compare the difference between the predicted value and the actual value and predict the infant mortality rate in Shanxi province from 2013 to 2015. Results ARIMA (1, 2, 0) fitted the time series of infant mortality in past time well. AR1 = -0.754, P <0.01, statistically significant. The AIC ) = 68.213, Schwab Bayes criterion (SBC) = 70.204, the model residual is white noise (P> 0.05). The mathematical function of the model is Yt = 0.067 + 1.246Yt-1 + 0.508Yt-2-0.754 Yt-3, using the model to predict 2013-2015 infant mortality rates were 4.77 ‰, 4.32 ‰, 3.96 ‰. Conclusions The ARIMA model can well fit the trend of infant mortality over time and can be used to predict future infant mortality rates in the short term.