Rough Set Data Mining Approach And Its Application Relative To Decision Problem

Data mining is a process in which hidden and previously unknown information with potential application value is extracted from database. Rough set theory is a mathematic tool for dealing with uncertain and ambiguous knowledge. It is very suitful for knowledge discovery and data mining, for the reason, it has some characteristics, such as the method of data driving, whithout any auxiliary information. In addition, it is easy to deal with discrete data and fuse with relevant database etc.Firstly, decision table reduction in the process of data mining, including attribute reduction and attribute value reduction, is studied in this thesis. About attribute reduction, a kind of algorithm based on discernibility matrix is improved through concluding the present reduction methods. Furthermore, the reduction algorithm based on ant colony optimization algorithm is put forward by combing rough set theory with intelligent algorithm. Secondly, about attribute value reduction, a kind of rapid method is also put forward in this thesis, which is based on ordering on the importance of attribute value. Then it mainly discusses attribute reduction of the incomplete information system and gives attribute reduction method based on genetic algorithm in the field of rough set model extension.Finally, a data-mining model based on rough set is designed in terms of the discussion in the former chapters. Also, the model is applied in mistake-preventing train system based on B/S structure for flight crew. The model system has excellent human-machine interface and general feature, in which the normal process about data mining is included. It is verified to have application value in practice.