摘要
在许多社会、经济、技术问题中 ,成分数据通常被用于研究作为整体的各部分的比率情况。本文结合对称的 logratio变换与偏最小二乘回归方法 ,提出一种成分数据回归建模的方法。该方法可以克服经典线性回归方法在成分数据建模中的诸多困难。并且由于所采用的数学变换具有对称性 ,因此变换以后的变量可以较好地反映原始变量的含义 ,便于建立一个容易解释的数学模型。本文还采用北京市三次产业就业需求预测作为案例 ,说明成分数据回归建模的工作过程和应用价值 。
In studies on the social, economic and technical problems, compositional data usually were used as the ratio development in each component of the whole. Based on the symmetrical logratio transformation and partial least squares regression, this paper proposed a simple linear regression modeling method. This method could overcome the difficulties that exist in the classic linear regression modeling. Furthermore, because of the symmetry of the mathematics transformation, the transformed variable can reflect the meaning of the original variable better, and so the mathematices model is easier to explain. Besides, this paper used the numerical forecast of employment in Beijing’s three industries as an example, illuminated the work process and application value of the regression modeling based on the compositional data, and made use of the result to verify the validity and rationality of this method.
出处
《系统工程》
CSCD
北大核心
2003年第2期102-106,共5页
Systems Engineering
基金
国家杰出青年科学基金资助项目 (70 12 5 0 0 3)
北京市自然科学基金资助项目 (90 0 2 0 0 2
