Incremental Algorithms For Temporal Data Mining Based On Rough Set Theory And Decision Tree

Temporal data mining is an important area on data mining and it has its own characteristic . When analyzing temporal data, you have to take time attribute in to consideration . The decision tree algorithm and rough set theory both are important method of classify in data mining. As the decision tree algorithm has advantage in making the rules extracted clear and natural on expression, knowledge acquired is easy to understand and speculate, while the rough set the rough set theory has advantage in processing data that is fuzzy and uncertain, considering the situation in the real world that the data are increasing everyday, we take advantage of the classic classify algorithm decision tree algorithm's main process procedure to deal with temporal data which is transferred. In the processing of creating a decision tree, we make good use of rough set theory to optimize the process of building the decision tree and drawing useful rules, then persent an increamental classify data mining algorithm.This paper firstly introduces the mathematics concept related to temporal data then introduces the method about transferring temporal data into the data without time attribute, proposing a new method to process this procedure . Then it introduces the basic theory about decision tree classification algorithm and rough set theory, analyzing the shortcoming on decision tree algorithm, especially on data mining for temporal data classification area . At last ,this article proposes Incremental Algorithms For Temporal Data Mining Based On Rough Set Theory And Decision Tree ,then it gives an example on stock market data mining, applying this algorithm.The main work of this paper proposes an improved the method of transferring temporal data into the data without time attribute . In the processing of building a tree , an improved the method of calculating information entropy , and suggested combining some time attributes as one combinative test attribute when necessary , and use rough set theory to cut off some useless rules, when processing increasing data ,we take consideration of this article's processing in temporal data , bring forward corresponding increment data handling method.To some extent, this paper is making attribution to calssify data mining area.