摘要
ID3算法是最经典的决策树算法,它以信息熵为基础,以信息增益划分测试属性。这种算法存在多值倾向问题,且存在局部最优问题。为了应对这些问题,本文引入肯德尔相关系数对信息增益进行改进修正。通过理论分析和实验验证,这种算法能较好的解决ID3算法的问题。
ID3 algorithm is the most classic decision tree algorithm.It is based on information entropy and divides test attributes by information gain.This algorithm has multi value tendency problem and local optimal problem.In order to solve these problems,this paper introduces Kendall correlation coefficient to improve the information gain.Through theoretical analysis and experimental verification,this algorithm can better solve the problem of ID3 algorithm.
出处
《科学技术创新》
2021年第22期21-22,共2页
Scientific and Technological Innovation
关键词
ID3算法
决策树
信息增益
多值倾向
局部最优
ID3 algorithm
Decision tree
Information gain
Multi value tendency
Local optimum
