Optimized algorithm of decision tree based on weighting factor
Dong Yue-hu · Journal of Jiangxi University of Science and Technology · 2015
Through the analysis of the issues of multivalue bias in the ID3 algorithm and subjectivity of the optimized traditional ID3 algorithm, an improved algorithm of decision tree based on weighting factor is put forward. The new algorithm introduces the weight factor that reflects the mutual relationship between the attributes. The ID3 algorithm is improved by redistricting the weight of attributes which has most values. The experiments on UCI data sets show that the optimization ID3 algorithm can overcome multivalue bias when the values of different attributes in data set are not the same. This algorithm not only improves the accuracy of average classification, but also reduces the number of average leaf nodes in the process of constructing a decision tree.