Intelligent Decision Research Based On Data Mining

With the rapid development of the science and technoledge, the scale of the database is becoming bigger and the quantities of data stored are rapidly increased. Therefore, The more new and more effective tools and technoledge are needed to mine all kinds of information resources in order to exert their application potential. The technoledge of knowledge discovery and data mining are produced on the background of application requirement and developed with the decision system. This paper recurred to the basic theory of rough set of data mining and under the direction of the frame of intelligent decision, the main theories include i) the different methods of data mining on the base of rough set are used to deal with typical decision system namely consistent decision system and inconsistent decision system in order to carry through data reduction and rule distilment; ii) In the environment of dynamic increment database, the methods of data reduction to deal with the original data and increment data are discussed in the consistent and inconsistent decision system;iii) The method of data mining of rough set is analysized to treat with the attributes with priority; iv) On the base of basic rough set theory, the data analysis methods of amalgamation of rough set theory; v) And also the pre- disposal method to database is analysize. The research and innovative results are as follows:1.The systems are rigourly divided into consistent and inconsistent decision system, and different data mining models are put forward in allusion to characterization of different systems.In consistent decision system, according to heuristic information from two angles the condition attributes are reduced. The first method used the indistinguish characterization of the knowledge and the mode of logical reasoning to condense the data in decision tables; The second method used the granularity of the knowledge and the mode of concept exaltation to condense the data in decision tables. In inconsistent decision system two improved algorithms namely decision conception conclusion and rough repetition groups are put forward to mine the classification rules with certainty reliability.2.In the dynamic increment database , data mining models of consistent and inconsistent decision system are formulated.In consistent decision system, using improved decision matrix, data mining model is put forward to treat with decision attribute with several values in order todirectly deal with increment data, which can obtain classification rules based on global data set; In inconsistent decision system the concept induce of inconsistent decision system is built with three parameters at the aid of combination of GDT with RS to generate the classification rules with probability condition.3. In allusion to the equivalence relation and priority relation of the condition attributes, the method of data mining of rough set is analysized to treat with the attributes with priority.On the base of extend rough set theory, this paper put forward the improved data mining algorithm with priority attributes based on the priority relation. Therefore, the classification precision of basic rough set and rough set with priority attributes reached unification and the classification rules by this model in the section are more curtail and rational.4. Build the extended models of the rough set and probability along with fuzzy set.Using the statistic characterization of data, the relevant knowledge reduction algorithm is put forward by combining the probability with classification rules; Using the characterization of fuzzy attributes, the decision system with subjection degree attribute is built by combing the rough set theory and fuzzy set theory, and the idea of distinguish matrix is induced to the concealed decision system to reduce data.5. Using the distributing characterization of data, a kind of data pre-process model with autocephaly field is put forward.This model used x2 statistic values to discretize condition attributes. Its coreide...