ID3 algorithm and C4.5 algorithm of decision tree are commonly used in the base classifier in random forest. In order to solve the problem that the basic classifier algorithm is biased to select multi-valued attributes and contains a lot of logarithmic operations, the attribute compatibility of rough set is introduced in this paper. Rough set is a mathematical tool for dealing with uncertain information. In this paper, the expression of information gain or information gain rate in the base classifier is reconstructed with compatibility, and the attribute with the highest compatibility is selected as the segmentation attribute to segment the data set. Finally, three UCI data sets are used to verify the accuracy of the algorithm. The experimental results show that the proposed algorithm has a higher accuracy than the traditional random forest algorithm.
[1]
Lu Liu,et al.
A new way to choose splitting attribute in ID3 algorithm
,
2017,
2017 IEEE 2nd Information Technology, Networking, Electronic and Automation Control Conference (ITNEC).
[2]
Aiko M. Hormann,et al.
Programs for Machine Learning. Part I
,
1962,
Inf. Control..
[3]
Huang Zhe-xue,et al.
A Brief Theoretical Overview of Random Forests
,
2013
.
[4]
Leo Breiman,et al.
Bagging Predictors
,
1996,
Machine Learning.
[5]
Leo Breiman,et al.
Bagging Predictors
,
1996,
Machine Learning.
[6]
J. Ross Quinlan,et al.
Induction of Decision Trees
,
1986,
Machine Learning.