Research And Application On Visualization Of The Process In Data Mining Interaction

As the development of the information processing technology, the data mining has drawn the concern of various circles, the main reason of this is there are a lot of data can be widely used and urgently need to be converted to useful information and knowledge.. But with its property of beyond comprehension for the result of data mining ,we must draw support from the concept and technology of data mining visualization to combine with the customer and the whole mining process, to advance the flexibility,the availability and the commutatively with the customer in the whole data mining process.This paper has made a summary of the key skills,the current research situation and the application prospects of the data mining and visualization, especially made an analysis for the visual model of decision tree. In the past data mining model, the application of visualization technology is limited, and the customer could neither understand the mining process, nor conduct the process of mining, still less understand well the mining result.In this paper, I made an analysis to the classificationID3 algorithms of decision tree, also made a deep analysis to its theory. The algorithm of ID3 bias toward the property of various access values, but a property with that is not certain to be a important property. We put up one algorithm to transform many kinds of problem to only two kinds of that. We state clearly its theory and carry on an analysis and comparison between the two kinds of experiment results of the two algorithms, and prove its viability.Designed and operated a decision tree visualization system, based on C++BUILDER as the development platform, we add some visualized part to the system to make the data visualization,inter-mining,model visualization and the foundation like that come true. To avoid the aimlessness of mining, we simplify the system according to source property which the entropy of information affect the decision by, and remove the property that is not important to the target to show the forwarded relations of the property by the methods of visualization. The decision tree visualization mainly includes leaf visualization and the tree visualization. If the customer is not satisfied with the result of mining, he can reecho the property to built model, and if satisfied, he can transform the decision tree to the decision rule. And finally calculate the model using the test data. The visualization model build by the system can provide the customer with more visible and easily understood and analyzed interface, with better mutual function, which is easy to conduct the mining process for the customer.