论文部分内容阅读
目的:探究分析灰色模型GM(1,1)、ARIMA及线性回归模型在孕产妇死亡率(MMR)中的拟合应用并比较模型的适用性及精确性,为妇幼保健工作提供科学依据。方法:以1991~2012年全国MMR为原始资料,运用SAS分析软件,分别采用灰色模型GM(1,1)、ARIMA及线性回归进行拟合比较分析并预测2013年MMR,比较预测效果。结果:灰色模型GM(1,1)和线性回归模型均通过模型检验建模成功,灰色模型GM(1,1)P=1.00,C=0.137 89,线性回归模型F=236.57,P<0.000 1;2013年MMR预测显示:线性回归模误差较大,误差的方差(MPE)为10.08,平均相对误差绝对值(MAPE)为13.69%。结论:ARIMA不适合拟合MMR,灰色模型GM(1,1)和线性回归模型均可用于MMR的拟合,但预测某一年份时需慎重。
OBJECTIVE: To explore and analyze the fitting application of gray model GM (1,1), ARIMA and linear regression model in maternal mortality (MMR) and to compare the applicability and accuracy of the model to provide a scientific basis for maternal and child health care. Methods: Based on the national MMR from 1991 to 2012, SAS software was used to compare and predict the 2013 MMR by using gray model GM (1,1), ARIMA and linear regression respectively. Results: Gray model GM (1,1) P = 1.00, C = 0.137 89, linear regression model F = 236.57, P <0.000 1 . The MMR forecast in 2013 showed that the error of linear regression model was larger, the error of variance (MPE) was 10.08 and the mean absolute relative error (MAPE) was 13.69%. Conclusion: ARIMA is not suitable for fitting MMR. Gray model GM (1,1) and linear regression model can be used to fit MMR, but caution should be exercised in predicting a year.