摘要
回归模型建立过程中如何选择合适的解释变量以及如何确定每个解释变量的函数形式至关重要,它直接影响模型的拟合能力和预测能力.论文在传统的变量选择的评价准则和变量选择方法的基础上,探讨了如何在模型中包含的变量不确定的情况下,准确有效的进行变量的选择,并在变量选择的同时进行变量的变换.旨在提高模型预测的精确度.
In the process of establishing a regression model, which explanatory variables and their function form to include in the model is crucial, it directly affects the model fitting ability and predictive power. Therefore on the basis of traditional variable selection methods, it discussed how to choose accurate and effective explanatory variables in the model, and discussed the method of simultaneous selection for variables and transformation, this aimed at improving the prediction accuracy of model.
出处
《河北工业大学学报》
CAS
北大核心
2009年第6期112-115,共4页
Journal of Hebei University of Technology
基金
河北省哲学社会科学规划研究项目(HB07BLJ003)
关键词
回归模型
变量选择
变量子集
Box.Cox变换
regression model
variables selection
variables subset
Box-Cox transformation
