Research On Extension Classification Knowledge Mining Based On Interval Manner

Extension data mining takes advantage of extension theories and manners to explore potential knowledges in enterprise databases. And then it offers helps in competition among businesses, decisions to produce and resolving contradictory problems. Currently, the main research content of extension data mining is the extension classification knowledge mining; However, there are also some disadvantages in the traditional extension classification knowledge mining. The first point is that the weight of integrated correlation function needs to be provided by the expert in the fields. That courses subjectivity and uncertainty in the approaches. Then the related researches about extension set and extension transformation still stay in the notion phase, lacking of the specific application manner.To solve the problems mentioned above, the paper adopts the manner based on intervals to mine the extension classification knowledge. The main works are as follows:First of all, the paper proposed a new method for weight assignment based on interval covers. It finds the best weights which can make the best classification effects from the data itself. Secondly, the paper proposed the extension transformation based on intervals mapping, offering a specific application manner for extension transformation. Meanwhile, the action scope is reduced to data belongs one class, and then improve the expression of extension set.Lastly, the traditional extension set which can only show the data of that could be classified into two classes is improved. When the data could be classified into more than two types, the method builds qualitative and quantitative change fields for the class which was changed by extension transformation and every other class. So the data of more classes can be displayed.After verified by UCI datasets, not only the improved correlation function is superior to some other classic classification algorithms on the accuracy, but also has advantage on the operating speed. Furthermore, the extension transformation proposed here and the improved extension set could offer solutions to the decision maker.