An Improved Fast Decision Tree Algorithm Based on Attribute Deviation
Liu Dan, Yue Zhang, Xin Sui, Zeng Yan, Wang Huan, Li Li · Frontiers in artificial intelligence and applications · 2017
This paper proposed a decision tree algorithm based on attribute deviation. The algorithm classified the target attribute values and replaced the “entropy” in the ID3 algorithm with a new classification rule. The new rules considered the deviation of the sample size for different attributes in the data set, completing a quick and detailed classification. At the same time, this classification rule solved the deviation caused by the decision tree bias selects multi-valued attributes. In this paper, a simulation experiment was conducted in UCI data set. The results showed that the attribute deviation algorithm has higher classification accuracy and shorter computation time than ID3 algorithm.