Based On The Decision Tree And Rough Set Classification

Data mining is the process that uses the analytical tools to extract some implicit, unknown, potentially useful information and knowledge from large, incomplete, noise, fuzzy and stochastic data sets, then the relationship among data could be established, thus forecast could be made with it, and assistance could be provided for the policy-maker. Decision tree is a common classification model, and it is used widely in data mining and other fields because that it can reflect characteristics of data directly, and it is efficient, fast, easy to understand; Rough sets theory is a effective method in dealing with imprecise, uncertain and incomplete data, and it has been applied successfully in a lot of domains, so, rough sets theory gets more attention from scholars in many countries.A research was made and some new ideas were put forward in this paper basing on decision tree and rough sets theory, the main contents of paper are listed as follow:In the first, a research was made basing on Decision tree, Rough sets theory and how it is represented by entropy, and the relationship was analyzed between upper-lower approximation and entropy. Because the effect of noise has not been though when used the entropy method to select the splitting attribute in famous ID3 algorithm, so it is sensitivity to noise, but variable precision rough set could restrain noise well, so, the ID3 was improved by combining this thought, after improving, it is fit for the data sets that have more noise, it accords with real life than before also.The second, the attribute reduction method in rough sets theory could reduce data set without affecting the ability of classification, so, a new algorithm based on the importance of attribute was put forward.