Research On Outlier Data Mining In High Dimensional Space

Outlier data mining can help people discover the true and unexpected information. Outlier data mining is a new task of data mining, and is widely used in daily life. At present, outlier data mining is a hotspot for the researchers of database, machine learning and statistics.Because of the characteristic of the distribution rule of the data in high dimension space, the traditional methods of outlier data mining can't work very well. To resolve this problem, a method of outlier data mining in high dimension space is presented in this paper, which uses rough set to reduce the attributes of the datasets and then detects outliers in the subspace using hypergraph model. The research tells us that the reduction of the attributes can save the storage space and using hypergraph model can detect the outlier data .The result of the experiment shows that this approach is effective and practical.This thesis is divided into six chapters. The fist chapter, Foreword, introduces the basic conception, methods, and category of data mining. The second chapter. Outlier Data Mining, is about the basic conception of outlier data mining and some ordinary outlier data mining methods. The third chapter, Rough Set Theory and Data Mining, introduces the basic theory of rough set and its relationship to data mining. The fourth chapter. Clustering, introduces some clustering methods and discusses the close relationship between clustering and outlier data mining. The fifth chapter. Research on Outlier Data Mining in High Dimension Space based on Rough Set and Hypergraph. presents an outlier data mining method in high dimension that uses rough set theory and hypergraph model. The last chapter is a conclusion of all the work in this paper and an expectation of further research work.