Decision Tree Based on Correlation Degree and Cost-sensitive Learning
Meng Guang-sheng · Science Technology and Engineering · 2013
Aiming at the shortcomings of the ID3 algorithms in decision tree generation,by introducing the attribute correlation degree and cost-sensitive learning,the method for generationg cost-sensitive decision tree based on attribute correlation degree is presented.The way reduces the attribute by rough set theory,and selects the splitting nodes by the attribute correlation degree and performance price ratio in generationg decision tree.The improved information gain is used for Generationg Cost-sensitive decision tree.The experimental results show that the method is superior in classification accuracy and node number than the usual algorithms in decision tree generation.