摘要
目的:立足卫生支出等趋势预测问题时间序列组合建模研究,借助算例比较验证。方法:根据政府卫生支出时序资料,将曲线拟合法和ARIMA法建模以后,考虑了残差修正组合法和线性加权组合法(残差平方和倒数法、关联度法、相关系数法、待定系数法和等权法)两大类组合途径,共建立六个组合模型。结果:修正指数曲线、ARIMA法拟合及预测效果不错;残差修正组合法实际算例不必要,其余五种线性组合模型效果优于任何单个方法模型。结论:曲线法、ARIMA法及六种组合模型对于趋势预测有应用意义。
Objective:Time series combination modeling research of trend prediction for example health expenditure as foothold,demonstration example was used to process comparison and verification.Methods:According to health expenditure time series data,curve-fitting method and ARIMA were introduced to model; residual modified model and linear weighting composite models were set up, such as residual sum of squares, grey correlation, correlation coefficient, undetermined coefficients and equal-weighted method. Fitting sequence and residual sequence were calculated, fitting performance improvement was discussed.Results:Amendment index curve model and ARIMA model were set up, fitting and prediction were good; combination technique was superior to respective method. Confirmed coefficients method was more effective. Conclusion:combination techniques based on curve and ARIMA methods existed application meaning for trend prediction problem.
出处
《中国卫生经济》
北大核心
2015年第8期56-58,共3页
Chinese Health Economics
基金
山东省高校社科项目(J14W21)
教育部人文社科项目(14YJAZH101,13YJAZH094)
山东省统计科研项目(2014-184)
国家自然科学基金项目(ZR2014-81473071)
潍坊市科技局项目(201301079)
山东省卫生计生委“十三五”规划课题
关键词
时间序列
组合建模
加权
卫生支出
趋势拟合
time series
combination modeling
weighting
health expenditure
trend fitting
