摘要
采用OneR算法计算出每个属性的准确率,准确率在一定程度上反映了属性与类标号属性的关联程度,使用准确率来修正信息增益公式以达到解决ID3算法中的多值依赖问题;另外,采用幂级数展开式对改进后的信息增益公式进行简化计算来提高运算效率;最后,采用实验验证了改进方案的可行性和优越性.
OneR algorithm is used to calculate the accuracy of each attribute,and the accuracy rate reflects the correlation between the attribute and the class label attribute to some extent.The accuracy is used to modify the information gain formula to solve the multi value dependency problem in ID3 algorithm.In addition,the power series expansion is used to simplify the calculation of the improved information gain formula to improve the operation efficiency.Finally,the feasibility and superiority of the improved scheme is verified.
作者
王利军
WANG Li-jun(Department of Information Engineering,Anhui Economics Management Institute,Hefei Anhui 230031,China)
出处
《菏泽学院学报》
2020年第5期15-19,30,共6页
Journal of Heze University
基金
安徽省高校自然科学重点项目(KJ2019A0965)。
关键词
准确率
算法改进
ID3算法
accuracy rate
algorithm improvement
ID3 algorithm
