计算机应用研究2016，Vol.33Issue(10)：3001-3006,6.DOI:10.3969/j.issn.1001-3695.2016.10.030
新型偏好敏感决策树算法
Novel preference sensitive decision tree algorithm
摘要
Abstract
In view of the fact that the existing decision tree models have not considered the preference behavior of decision makers to results during the classification process,so that it cannot predict problems which have obvious preference problem very well.This paper proposed a new PSDT algorithm.The algorithm introduced the preference degree and preference cost.It constructed a new attribute selection factors (ASF)by taking attribute information and effective preference into account.Fur-ther more,the PSDT algorithm had the class label distribution rule of node based on effective preference.Finally,it generated a preference sensitive decision tree which had the best preference degree by adjusting the preference degree adaptively.The theoretical analysis and experimental results show that the algorithm can not only achieve the high precision prediction of pref-erence class,but also ensure the decision tree has good overall accuracy.And it has high validity and practicability.What is more,it can solve the decision problems in preference sensitive environment in a good method.关键词
决策树/偏好敏感/偏好度/属性选择/代价敏感/分类Key words
decision tree/preference sensitive/preference degree/attribute selection/cost sensitive/classification分类
信息技术与安全科学引用本文复制引用
周美琴,徐章艳,陈诗旭,李艳红,马顺,展雪梅..新型偏好敏感决策树算法[J].计算机应用研究,2016,33(10):3001-3006,6.基金项目
国家自然科学基金资助项目（61462010，61363036）；广西多源信息挖掘与安全重点实验室主任基金资助项目；桂林市科技计划项目；八桂学者团队资助项目 （）
