摘要
本文根据婴儿死亡率随人均GDP的动态变化规律筛选最佳验证模型,验证了通过时间序列模型重新构建的1952—1980年婴儿死亡率和调整校正的1981—1990年婴儿死亡率。结果表明幂函数形式为相对较好的验证模型,验证模型与时间序列模型的预测结果与婴儿死亡率历史数据比较,变异程度无显著差异,且预测结果与建立国家儿童死亡监测网络后的婴儿死亡率形成了平稳性过渡。本文认为以时间序列模型重新构建和调整校正的婴儿死亡率比较可靠,更加接近当时的实际水平。
According to the infant mortality dynamic changes with GDP per capita, we screened the best verification model. Then we verified reconstructed infant mortality rate （1952--1980） and adjusted infant mortality rate （1981-- 1990） by the time series model. The results showed that the power function model was relatively good validation model, the fitting precision of which was poorer than the time series prediction model. When the predictive results of validation model and the time series model were compared with history infant mortality, there was no significant difference between them. And the predictive results coincided with infant mortality monitored after establishment of a national monitoring network. We thought that the reconstructed and adjusted infant mortality rate by time series was more reliable and closer to the actual level.
出处
《统计研究》
CSSCI
北大核心
2015年第12期84-87,共4页
Statistical Research
关键词
婴儿死亡率
验证
动态模型
预测
重新构建
调整校正
Infant Mortality Rate
Validation
Dynamic Model
Forecast
Reconstruction
Adjustment
