Data Mining Based On Rough Set Reduction Algorithm With Applications

With the rapid development of information technology, the application of data base become more and more widely in different areas, and the previous processing methods of data base can't meet the needs any more. Data mining is a new technology of data processing to meet the needs. It can take out the potential useful ones amony a great deal of information automatically, and test and verify them, returning the most useful results to the users without any explicit hypotheses. This technology is widely used in more and more areas; it has a bright future in application.The rough set theory solves the problem of a great deal of redundant data that influence our decision by the rule of decision and the progress of reasoning. It can not only decrease the data amount (data concentration), but also produce decision rules to take out the efficient models among data under the condition of not changing the data information. Unlike other theories to solving inexplicit problems, rough set theory doesn't need the previous verify information except the demanding data set to processing.In this paper, firstly, the study on the progress of data mining (the data processing in advance,,reduction and the rules) based on rough set was conducted. And the key topic is attribute reduction with which a further study was conducted, analyzing the algorithm of attribute reduction that is popular nowadays, and pointing out the existing problems.Secondly, an algorithm based on DMI was put forwarded that can not only decrease the complexity of time and space, but also get the minimize attribute reduction. The main contents are the following:1. The analysis on discernibility matrix: showing this algorithm is incorrect to incompatible resolution lists with examples, and putting forwarding an improving algorithm of attribute reduction suiting for compatible resolution lists and incompatible ones.2. Aiming at the weakness of current attribute reduction that have to read all data items of resolution lists, complexing the time and space, an improving algorithm was put forwarded to simplify the time and space by only reading resolution lists once.3. Showing the importance of attribute value reduction to data reduction, practically realizing the simplicity of resolution lists. And putting forwarded the improved algorithm of value attribute reduction that decreases the complexity of time greatly.4. Realizing the unify of attribute reduction and attribute value reduction by fusing the three improved thoughts of reduction together, and that is the real algorithm of reduction in deed. Considering the significant of the whole resolution type, the whole reduction progress of algorithm was conducted aiming at all targets. Comparing with the current methods that process separately attributes reduction and attribute value reduction with different algorithm, this algorithm decreased the space complexity greatly. Therefore, this algorithm has great practice significant and bright future of application.5. With the real consideration of the meaning of the whole class of decision to the reduction , the whole process of algorithm focuses on all the objects at the same time which avoid the mistakes in gaining partial data. The experiences forwards the UCI database has testified the improved algorithm's advanced and practicable.And lastly, the improved reduction algorithm was applied to the project studying on the innovation of agricultural water price and farmer's ability to offer. It is the first time to put forward...