Heuristic Mode Research and Application of Decision Tree Algorithm

Abstract

Decision tree, as an important classification algorithm in data mining, has been successfully applied in many fields. In this paper, based on the analysis of the essential characteristics of decision tree algorithm, we give a leaf criterion for multi-decision values of decision attribute, and establish a mathematical model for the selection for expanded attributes; also we give a concrete model based on quasi-linear function (denoted by QASM). Finally, we compare and analyze the performance of QASM combining with ID3 algorithm through an example. The results show that QASM can not only effectively merge the decision consciousness into decision-making process in a quantitative way, but also the computational complexity is lower than that of ID3 algorithm.