成分数据偏最小二乘 Logistic回归模型及其应用

成分数据由于存在定和约束和多重共线性，在进行传统的logistic回归建模时遇到了困难。本文在对称logratio变换和偏最小二乘logistic回归技术的基础上，提出了成分数据偏最小二乘logistic回归模型，较好地解决了成分数据的logistic回归建模问题。应用此方法，本文研究了中国三次产业就业结构与人民生活水平之间的关系。结果表明，该模型在有关结构问题的logistic回归建模中具有很好的适用性。

PLS Logistic Regression on Compositional Data and Its Application

Ordinary logistic regression is not available when the independent variables are compositional data, which belong to a special data type with multicollinearity and constraint of constant sum. Based on the techniques of symmetrical logratio transformation and PLS logistic regression, this paper proposes a new method of PLS logistic regression on compositional data. Applied this method, a case of the structure of employment in China has been studied. The result indicated that the approach presented is applicable to solve the problems of logistic regression modeling related to structure problems.