关键词: 数据挖掘, 分类, 不完备信息系统, 贝叶斯算法

Abstract:

Classification is an important data mining function. Featuring high accuracy, Bayesian algo-rithm is one of the classical classification algorithms. However, Bayesian algorithm requires that the datato be processed is a complete information system, and the algorithm cannot be directly applied to incom-plete information systems. In order to establish a classification model for incomplete information systems.this paper improves the DBCI classifier and constructs the lBCI classifier, which does not need to supple-ment the data in advance. The classifier avoids the low accuracy of classification model caused by datacomplement. Both the theory and the experimental results prove that the classifier has higher accuracythan the DBCI classifier, and extends the application scope of Bayesian algorithm.

Key words: data mining, classification, incomplete information systems, Bayesian algorithm