Research On Data Pretreatment And Decision Table Reduction Based On Rough Set Theory

In 1982, the Polish scholar Z.Pawlak proposed the Rough Set Theory which is a tool to deal with the imprecise, inconsistent, incomplete information system. The Rough Set Theory can handle such problems as data reduction, data mining, the evaluation of attribute importance, the formation of decision algorithm etc.Dealing with incomplete systems is an important part of data pretreatment. People often face incomplete information systems when acquiring acknowledge, which means that some attribute values are missing in the information systems. Considering this point, this paper discusses the methods dealing with the incomplete systems. Many researchers have studied how to discretize a continuous attribute and suggested some methods such as S method, H method, L method, and so on. These methods often result in too many equivalence classes to be included in a database. In order to overcome this drawback, this paper proposes a new method based on the Rough Set Theory for data generalization. It is shown that the pattern discovered by this method has a higher precision and more powerful interpreting ability. Knowledge reduction is an essential part of the Rough Set Theory, which includes both attributes reduction and attribute values reduction. It demands keeping the ability of assorting and deciding unchanged while deleting the irrelevant or unimportant knowledge. The paper develops an algorithm for decision table reduction. Through constructing a Discernibility Table, the algorithm can extract the most important attribute and its relevant attribute values, but does not have to examine the objects one by one. The classical Rough Set Theory requires a complete or correct system when the system involves assorting. But in fact, it is sometimes unavoidable for the system to contain noise data. In order to resist the noise data, Variable Precision Rough Set Theory was proposed. In this paper, the Variable Precision Rough Set Theory is applied to judge the development degree of a region. It can resist the interference of the noise and get simple rules, but does not change the ability of assorting.