具有二次函数的多准则层次分类方法

发布时间：2018-04-12 19:03

本文选题：多准则决策 + 分类　；参考：《系统工程与电子技术》2006年11期

【摘要】：提出了一种集成函数是二次函数且有训练集的多准则层次分类决策方法。该方法利用决策者对训练集的分类结果(属于最高分类和不属于最高分类)构建非线性规划模型,然后采用一系列处理方法将非线性规划模型转换成线性规划模型,求解线性规划,得到训练集中各方案准则值的偏好值和相应参数,通过线性插值或样条插值得到方案在各准则下的偏好值,并计算方案集中方案的与该分类一致性指标值和与低于该分类一致性指标值的差,以确定方案是否属于该分类,然后对不属于最高分类的训练集中的方案进行分类,并构建模型。继续上述过程,直到方案集中所有方案均进行分类为止。最后实例说明该方法的有效性和可行性。
[Abstract]:In this paper, a multi-criteria hierarchical classification decision method with training set is presented, in which the integration function is a quadratic function.In this method, a nonlinear programming model is constructed by using the classifying results of the training set (belonging to the highest classification and not the highest classification), and then a series of processing methods are used to transform the nonlinear programming model into a linear programming model.By solving linear programming, the preference values and corresponding parameters of each scheme criterion value in the training set are obtained, and the preference values of the scheme under each criterion are obtained by linear interpolation or spline interpolation.The difference between the consistency index value of the scheme set scheme and the consistency index value of the classification is calculated to determine whether the scheme belongs to the classification or not, and then the scheme of the training set that does not belong to the highest classification is classified.And build the model.Continue the above process until all schemes in the scheme are classified.Finally, an example is given to illustrate the effectiveness and feasibility of the method.
【作者单位】：中南大学商学院
【基金】：湖南省社会科学基金资助课题(05YB74)
【分类号】：C934

【参考文献】