Algorithm Improvement And Application Based On Attribute Significance

As a new technology for processing large amounts of data, data mining has the ability to analyze and explore the knowledge implied in the data. Especially, the rough set theory plays a very active role among these technologies, and the classical rough set theory is widely used in signal processing. It has the advantage of dealing with incomplete and inaccurate information without the utilization of priori knowledge, and explores the implied knowledge and the potential rules in the process. Therefore, data mining based on rough set has attracted study interes.This dissertation mainly focuses on the attribute reduction, as well as its application in the analysis of the affecting factors in the students overall performance.Firstly, a review of the classical rough set theory is presented, mainly including its background, significance and development. A discussion about the research on the students' performance and the common method is made. The knowledge reduction is studied from a perspective of information theory other than theory of algebra, it is beneficial to handling with the uncertain problems.Secondly, in order to find out an effective discrete method, a comparasion is made among the common discretization methods. In this way, a effective discrete method well meeting the actual situation is explored.Finally, an improved attribute reduction algonthm based on attributes significance is put forward, and the affecting factors of the overall scores is analyzed using the algonthm In accordance with the decision table, the construction thought, the procedure and the reduction algorithm is described. The reduction and the analysis are made, and the results agree well with the software computing result. The generated rules is of great value in improving the students overall performance in the school education and teaching.The final results of this dissertation is to design a attribute reduct improved algorithm of attribute significance. Using this algorithm we can reduce the complexity of the original algorithm. Finally, we can acquire the optimal decision making rules from the decision making system.