基于偏好不一致熵的有序决策

发布时间：2018-06-10 18:36

本文选题：有序决策 + 偏好不一致熵　；参考：《计算机应用》2017年03期

【摘要】：针对多规则有序决策系统中的偏好决策问题,根据有序决策的偏好不一致特性,提出了一种基于偏好不一致熵的偏好决策方法。首先,定义了样本的偏好不一致熵(PIEO),用来度量特定样本相对于样本集的偏好不一致程度;然后,根据偏好决策中不同属性对决策的重要性不同的特点,提出了一种加权的样本偏好不一致熵,并结合属性偏好不一致熵在度量属性重要性方面的能力,给出了一种基于属性偏好不一致熵的权值的计算方法;最后,提出了一种基于样本偏好不一致熵的偏好决策算法。采用Pasture Production和Squalsh两个数据集进行仿真实验,基于全局偏好不一致熵分类后,各属性的偏好不一致熵普遍比基于向上和向下偏好不一致熵分类后的熵值小,而且更接近原始决策的偏好不一致熵,这说明基于全局偏好不一致熵的分类比其他两种情况的分类效果好。分类偏离度最小低至0.128 2,这说明分类的结果比较接近原始决策。
[Abstract]:In order to solve the preference decision problem in multi-rule ordered decision system, a preference decision making method based on preference inconsistent entropy is proposed according to the preference inconsistency of ordered decision. Firstly, the preference inconsistency entropy of samples is defined to measure the degree of preference inconsistency relative to the sample set, and then, according to the characteristics of the importance of different attributes in preference decision, In this paper, a weighted sample preference inconsistent entropy is proposed, and combining the ability of attribute preference inconsistent entropy to measure attribute importance, a weight calculation method based on attribute preference inconsistent entropy is presented. A preference decision algorithm based on sample preference inconsistent entropy is proposed. By using two data sets, Pasture production and Squalsh, the entropy of preference inconsistency is generally smaller than that based on upward and downward preference inconsistency entropy classification based on global preference inconsistent entropy classification. Moreover, it is closer to the preference inconsistent entropy of the original decision, which shows that the classification based on the global preference inconsistent entropy is better than the other two cases. The minimum deviation of classification is as low as 0.128 2, which indicates that the classification results are close to the original decision.
【作者单位】：电子科技大学信息与软件工程学院;西华师范大学计算机学院;
【基金】：四川省教育厅自然科学基金重点资助项目(12ZA178) 四川省重大项目支撑计划项目(2015GZ0102) 四川省可视计算和虚拟现实重点实验室建设基金资助项目(KJ201406)~~
【分类号】：O225

【相似文献】