基于预测有效度成分数据处理方法组合预测
|
摘要
为提高成分数据时序预测准确性,提出一种以二阶预测有效性作标准的多种数据处理方法的组合预测。选择成分数据的多种数据转化方法,将有约束时序用对数比,中心对数,超球面变换方法转换成无约束时序后,利用ARIMA—ANN模型对转换后无约束时序预测,对结果做反变换,恢复为成分数据得单项预测结果。最后对得到的单项预测结果进行基于二阶预测有效度的加权几何平均组合,得到相对最优的组合预测结果。
In order to improve the accuracy of time series prediction of component data, a combined prediction of multiple data processing methods based on second-order prediction validity is proposed. Selecting a variety of data transformation methods for component data,after the constrained time series is transformed into the unconstrained time series by the logarithmic ratio, the central logarithm and the hypersphere transformation method, the ARIMA-ANN model is used to predict the unconstrained time series after the transformation, and the result is inversely transformed to restore the component data to a single prediction. result Finally, the weighted geometric mean combination based on the second-order prediction validity is obtained for the obtained single prediction result, and the relatively optimal combined prediction result is obtained.
In order to improve the accuracy of time series prediction of component data, a combined prediction of multiple data processing methods based on second-order prediction validity is proposed. Selecting a variety of data transformation methods for component data,after the constrained time series is transformed into the unconstrained time series by the logarithmic ratio, the central logarithm and the hypersphere transformation method, the ARIMA-ANN model is used to predict the unconstrained time series after the transformation, and the result is inversely transformed to restore the component data to a single prediction. result Finally, the weighted geometric mean combination based on the second-order prediction validity is obtained for the obtained single prediction result, and the relatively optimal combined prediction result is obtained.
引文
[1]Reyment R A.The statistical analysis of compositional data[J].Chemometrics&Intelligent Laboratory Systems,1988,3(4):254-256.
[2]Piepel D,Gregory F.The Statistical Analysis of Compositional Data[J].Technometrics,1988,30(1):120-121.
[3]王惠文,刘强.成分数据预测模型及其在中国产业结构趋势分析中的应用[J].管理评论,2002(5):27-29.
[4]Egozcue J J,Pawlowskyglahn V,Mateufigueras G.Isometric Logratio Transformations for Compositional Data Analysis[J].Mathematical Geology,2003,35(3):279-300.
[5]宁自军.成分数据的预测方法与应用[J].统计与决策,2001(6):6-7.
[6]Bracci P,Bull S,Grynpas M D.Analysis of compositional bone density data using log ratio transformations[J].Biometrics,1998,54(1):337-349.
[7]刘勇.ARIMA模型在我国能源消费预测中的应用[J].经济经纬,2007,26(5):11-13.
[8]陈华友.基于预测有效度的组合预测模型研究[J].预测,2001,20(3):72-73.
[2]Piepel D,Gregory F.The Statistical Analysis of Compositional Data[J].Technometrics,1988,30(1):120-121.
[3]王惠文,刘强.成分数据预测模型及其在中国产业结构趋势分析中的应用[J].管理评论,2002(5):27-29.
[4]Egozcue J J,Pawlowskyglahn V,Mateufigueras G.Isometric Logratio Transformations for Compositional Data Analysis[J].Mathematical Geology,2003,35(3):279-300.
[5]宁自军.成分数据的预测方法与应用[J].统计与决策,2001(6):6-7.
[6]Bracci P,Bull S,Grynpas M D.Analysis of compositional bone density data using log ratio transformations[J].Biometrics,1998,54(1):337-349.
[7]刘勇.ARIMA模型在我国能源消费预测中的应用[J].经济经纬,2007,26(5):11-13.
[8]陈华友.基于预测有效度的组合预测模型研究[J].预测,2001,20(3):72-73.