文献类型：期刊文献

中文题名：可拓逻辑和可拓思维对决策树算法的分析与优化

英文题名：Analysis and Improvement of Decision Trees with Extension Logic and Extension Thinking Mode

作者：朱弘扬[1];丁怡[2];柴华金[1];李升[1]

机构：[1]广东海洋大学数学与计算机学院;[2]广东海洋大学机械与动力工程学院,广东湛江524088

年份：2019

卷号：36

期号：2

起止页码：31

中文期刊名：广东工业大学学报

外文期刊名：Journal of Guangdong University of Technology

基金：广东省教育厅创新强校项目(2016WQNCX052)

语种：中文

中文关键词：可拓逻辑;可拓思维;决策树算法;矛盾问题

外文关键词：extension logic;extension thinking mode;decision trees;contradictory problems

中文摘要：为从理论层次上深度解析决策树分类算法的逻辑可行性,根据可拓理论中可拓逻辑和可拓思维的全新视角,对构建决策树过程中节点的选择、规则提取和预测等步骤,进行理论上的分析和评价.以可拓思维中的菱形思维模式来分析决策树算法中节点的选择,以可拓逻辑中基元变换理论来评价决策树算法的规则提取,以可拓逻辑中的基元发散规则来解释决策树算法的预测步骤,在验证决策树算法各个步骤符合可拓理论处理矛盾问题的思维模式的同时,也对决策树算法的分类结果建立了基于可拓理论的评价体系.

外文摘要：To analyse the logical feasibility of Decision Trees algorithm in theory, the choice of the note, the extraction of rules and the prediction in decision tree are studied from a novel perspective with extension logic and extension thinking modes. By analysing the processes of note-choice and rule-building in decision tree based on the principles of rhombus thought and element transformation, and evaluating the prediction of decision tree based on element divergence in Extenics, it’s found that the logic of decision tree conforms to the concept of Extenics which is used to solve contradictory problems. Besides, an evaluation system for the classification result of decision tree is built based on the Extenics theory.