Research And Implementation Of Visual Data Mining Technology

With a strong information processing ability, association rules mining can find interesting and relative relations among large amounts of data. However, current data mining techniques which support association rules mining can not interact with users, which make the result hard to identify.In this thesis, the integration and application of data mining and visualization are firstly analyzed, and then the feasibilities of applying visualization in each phase of data mining are studied. On the basis of these study an association rules based all process data mining visualization method is proposed, and the system of this method is realized. In this system the source data is visualized by the Principal Component Analysis dimension reducing method and two-dimensional scatter plot method, and the mining result is visualized three-dimensionally with Java 3D. What's more, an integrated friendly visual interface is designed to guide user to complete the association rules data mining tasks.Finally, experiment on a web trading database shows that the visual data mining system is feasible and effective.