在多准则决策任务中,属性和决策是有序的.传统的特征评价指标在这类任务上应用效果受到一些限制,获取有效的特征需要考虑特征值的有序特性.为了辅助管理决策,分析了三种有序的特征评价指标,为进一步解决复杂的决策问题提供依据.通过实验,说明不同评价指标的性能差异,以及在有序决策树算法中应用的效果和鲁棒性,验证所有指标在不同任务上的可行性和有效性.%In multi-criteria decision tasks,attributes and decisions may be ordinal.The traditional feature evaluation index is restricted in some tasks for application effect that the valid feature should be obtained considering the orderly characteristics.In order to assist the management decision, three orderly evaluation indexes are analyzed, which provides the basis for further solving complex decision problems.According to the experimental results,the performance difference of different evaluation indicators and the effect applied in the classification algorithm are explained, and the feasibility and validity of all the indicators on different tasks are verified.
展开▼