An Improved Self-mapping Space Decision Tree Algorithm
ZHANG Shu-yu, ZHU Zhong-ying(Dept. of Automation, Shanghai Jiaotong Univ., Shanghai 200030, China)
The decision tree is very useful in building knowledge-based expert system, and it is also a powerful method in (spatial) data mining. But the current decision tree induction methods do not deal with vagueness and ambiguity associated with human thinking and perception very well. This paper presented a self-mapping space(SMS) model for knowledge representation and uncertainty handling. Compared with classical induction method, the SMS integrates the fuzziness and randomness of linguistic terms in a better way. A new kind of decision tree based on SMS model (SMS Decision Tree, denoted as SMS-DT) was developed, and the detailed induction method was presented. The method associates naturally with human thinking and perception. In a practical spatial classification problem, the SMS method shows the benefits in effectiveness and flexibility.
【CateGory Index】： TP18