摘要
目的本文以医院住院人数序列为例阐述春节效应调整的方法。方法根据每段效应期内落入某月份的天数占该段效应期总天数的比例确定春节效应回归因子,采用regARIMA模型进行春节效应的调整。结果调整后序列的年度均值近似,频域消除了年度周期峰值且其他周期基本不变,调整后序列的预测效果明显优于调整前序列。结论采用regARIMA模型进行春节效应调整是必要的,其调整效果令人满意。
Objective Using the number of admission sequence to introduce the adjustment method of the Spring Festival effect.Methods Take daylength falling into each effect period in each month above the total number of days in this effect period as a proportion.According to the proportion to determine regression factors put them into regARIMA model,and perform the Spring Festival effect adjustments.Results Annual average of adjusted sequence is similar to unadjusted sequence.In frequency domain,the adjusted sequence eliminates the annual periodicity and remains other periodicity peaks almost unchanged.The forecast effectiveness of adjusted sequence is better than the unadjusted sequence.Conclusion Using regARIMA model to do the adjustment of the Spring Festival effect is necessary,with satisfactory forecasting effectiveness.
出处
《中国卫生统计》
CSCD
北大核心
2010年第5期473-476,共4页
Chinese Journal of Health Statistics
基金
2008年国家自然科学基金资助(30872182)
