An Investigation of Coal Demand in China Based on the Variable Weight Combination Forecasting Model

Variable weight combination forecasting combines individual forecasting models after giving them proper weights at each time point. Weight is the type of function that changes with forecast time. A relatively rational description of the system can be proposed with the forecasting method, which is of higher precision and better stability. Two individual forecasting models, grey system forecasting and multiple regression forecasting, were generated based on the historical data and influencing factors of coal demand in China from 1981 to 2008. According to the theory of combination forecasting, the variable weight combination forecasting model was formulated to forecast coal demand in China for the next 12 years.

医院季度收入预测的比较研究

目的 探讨具有季度波动和长期趋势特征事物预测问题的方法论,为卫生领域同类预测问题提供定量决策方法论依据。方法 以某所医院季度收入预测作为实证算例载体,分别用乘积季节SARIMA法、X11-ARIMA法、X11-曲线函数外推法建立简单组合模型以及二次加权组合模型。结果 乘积季节SARIMA模型、X11-ARIMA模型、X11-二次多项式的平均相对误差分别为1.28%、0.96%、1.11%。对以上简单组合模型继续进行深度加权组合,预测误差为0.70%,模型预测性能更优。结论 SARIMA模型、X11-ARIMA模型、X11-二次多项式以及二次组合模型对于局部周期性波动、全局稳定性趋势的时序数据资料拟合效果不错;对于卫生管理领域预测问题有适用性,鉴于多种匹配方法原理不同、信息利用差异大,基于多种方法的二次组合模型取得更优性能。